Thermodynamic and parametric modeling in the refining of

high carbon ferrochromium alloys using manually operated

AODs

MSc (5050) RESEARCH PROJECT

Prepared by

Kelvin Mukuku (782232)

Submitted to

School of Chemical and Metallurgical Engineering Faculty of Engineering and the Built

Environment University of the Witwatersrand Johannesburg South Africa

Supervisor Dr E Matinde

18 July 2017

DECLARATION

I declare that this report is my own unaided work I am submitting it in partial fulfilment of the

requirements of Master of Science in Metallurgical Engineering at the University of

Witwatersrand Johannesburg

It has not been submitted before for any degree or examination to any other university

helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip

(Signature of Candidate)

ii

DEDICATION

I dedicate this work to my family they were patient with me whilst I dedicated more

time to my school work

iii

ACKNOWLEDGEMENTS

I would like to extend my thanks to the following people

1 My Supervisor Dr E Matinde whose guidance was the source of my strength even

when I thought I could not go on

2 Mr Jacques Beylefeld of XRAM Technologies who took his time to explain to me

some complicated thermodynamic principles

3 Mr Mark Verwayen a friend and brother who introduced me to AOD technology

iv

ABSTRACT

This study and the work done involves investigating the effects of different parameters on the

decarburization process of high carbon ferrochromium melts to produce medium carbon

ferrochrome and takes into account the manipulation of the different parameters and

thermodynamic models based on actual plant data Process plant data was collected from a

typical plant producing medium carbon ferrochrome alloys using AODs The molten alloy was

tapped from the EAF and charged into the AOD for decarburization using oxygen and nitrogen

gas mixtures The gases were blown into the converter through the bottom tuyeres Metal and

slag samples and temperature measurements were taken throughout the duration of each heat

The decarburization process was split into two main intervals namely first stage blow (where

carbon content in the metal bath is between 2-8 wt C) and second stage blow (carbon mass

below 2 wt ) The first and second blow stages were differentiated by the gas flow rates

whereby the first stage was signified by gas flow ratio of 21 (O2N2) whilst the stage blow had

11 ratio of oxygen and nitrogen respectively

The effect of Cr mass on carbon activity and how it relates to rate of decarburization was

investigated and the results indicated that an increase in Cr 6654 ndash 705 wt reduced carbon

activity in the metal bath from 0336 ndash 0511 for the first blowing stage For the second blowing

stage the increase in Cr mass of 6722 ndash 7165 wt resulted in an increase in C activity from

0336 ndash 057 The trend showed that an increase in chromium composition resulted in a decrease

in carbon activity and the same increase in Cr mass resulted in reduced carbon solubility

Based on the plant data it was observed that the rate of decarburization was time dependent that

is the longer the decarburization time interval the better the carbon removal from the metal

bath An interesting observation was that the change in carbon mass percent from the initial

composition to the final (ΔC) decreased from 1018 ndash 837 wt with the increase in CrC

ratio from 837 ndash 1018 This effect was attributed to the chromium affinity for carbon and the

fact that an increase in chromium content in the bath was seen to reduce activity of carbon It

was also observed that the effect of the CrC ratio was more significant in the first stage of the

blowing process compared to the second blowing stage A mass and energy balance model was

constructed for the process under study to predict composition of the metal bath at any time

interval under specified plant conditions and parameters The model was used to predict the

outcome of the process by manipulating certain parameters to achieve a set target By keeping

v

the gas flow rates blowing times gas ratios and initial metal bath temperature unchanged the

effect of initial temperature on decarburization in the converter was investigated The results

showed that the carbon end point with these parameters fixed decreased with increasing initial

temperature and this was supported by literature The partial pressure of oxygen was observed

to increase with decrease in C mass between the first and second blow stages For the second

stage blow the partial pressure changed from 55210-12

ndash 2110-10

and carbon mass

increased from 0754 ndash 299 wt A carbon mass of 787 had an oxygen partial pressure of

45110-13

whilst a lower carbon content of 153 wt had an oxygen partial pressure of

80610-11

The CO partial pressure however increased with increase in carbon composition in

the metal bath

When the oxygen flow rate increased a corresponding increase in the carbon removed (ΔC)

was observed For the first stage of the blowing process an increase in oxygen flow rate from

38867 ndash 6665Nm3 resulted in an increase in carbon removed from 506 ndash 728 wt The

second blowing stage had lower oxygen flow rates because of the carbon levels remaining in the

metal bath were around +- 2 wt In this stage oxygen flow rates increased from 125 ndash 28667

Nm3 and carbon removed (ΔC) from 016 ndash 2093 wt The slag showed that an increase in

basicity resulted in an increase in Cr2O3 in the slag As the basicity increased from 0478 ndash

1281 this resulted in an increase in Cr2O3 increase from 026 ndash 068 Nitrogen solubility in the

metal bath was investigated and it was observed that it increased with increasing Cr mass The

increase in nitrogen solubility with increasing Cr mass was independent of the nitrogen partial

pressures

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

DECLARATION

I declare that this report is my own unaided work I am submitting it in partial fulfilment of the

requirements of Master of Science in Metallurgical Engineering at the University of

Witwatersrand Johannesburg

It has not been submitted before for any degree or examination to any other university

helliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphelliphellip

(Signature of Candidate)

ii

DEDICATION

I dedicate this work to my family they were patient with me whilst I dedicated more

time to my school work

iii

ACKNOWLEDGEMENTS

I would like to extend my thanks to the following people

1 My Supervisor Dr E Matinde whose guidance was the source of my strength even

when I thought I could not go on

2 Mr Jacques Beylefeld of XRAM Technologies who took his time to explain to me

some complicated thermodynamic principles

3 Mr Mark Verwayen a friend and brother who introduced me to AOD technology

iv

ABSTRACT

This study and the work done involves investigating the effects of different parameters on the

decarburization process of high carbon ferrochromium melts to produce medium carbon

ferrochrome and takes into account the manipulation of the different parameters and

thermodynamic models based on actual plant data Process plant data was collected from a

typical plant producing medium carbon ferrochrome alloys using AODs The molten alloy was

tapped from the EAF and charged into the AOD for decarburization using oxygen and nitrogen

gas mixtures The gases were blown into the converter through the bottom tuyeres Metal and

slag samples and temperature measurements were taken throughout the duration of each heat

The decarburization process was split into two main intervals namely first stage blow (where

carbon content in the metal bath is between 2-8 wt C) and second stage blow (carbon mass

below 2 wt ) The first and second blow stages were differentiated by the gas flow rates

whereby the first stage was signified by gas flow ratio of 21 (O2N2) whilst the stage blow had

11 ratio of oxygen and nitrogen respectively

The effect of Cr mass on carbon activity and how it relates to rate of decarburization was

investigated and the results indicated that an increase in Cr 6654 ndash 705 wt reduced carbon

activity in the metal bath from 0336 ndash 0511 for the first blowing stage For the second blowing

stage the increase in Cr mass of 6722 ndash 7165 wt resulted in an increase in C activity from

0336 ndash 057 The trend showed that an increase in chromium composition resulted in a decrease

in carbon activity and the same increase in Cr mass resulted in reduced carbon solubility

Based on the plant data it was observed that the rate of decarburization was time dependent that

is the longer the decarburization time interval the better the carbon removal from the metal

bath An interesting observation was that the change in carbon mass percent from the initial

composition to the final (ΔC) decreased from 1018 ndash 837 wt with the increase in CrC

ratio from 837 ndash 1018 This effect was attributed to the chromium affinity for carbon and the

fact that an increase in chromium content in the bath was seen to reduce activity of carbon It

was also observed that the effect of the CrC ratio was more significant in the first stage of the

blowing process compared to the second blowing stage A mass and energy balance model was

constructed for the process under study to predict composition of the metal bath at any time

interval under specified plant conditions and parameters The model was used to predict the

outcome of the process by manipulating certain parameters to achieve a set target By keeping

v

the gas flow rates blowing times gas ratios and initial metal bath temperature unchanged the

effect of initial temperature on decarburization in the converter was investigated The results

showed that the carbon end point with these parameters fixed decreased with increasing initial

temperature and this was supported by literature The partial pressure of oxygen was observed

to increase with decrease in C mass between the first and second blow stages For the second

stage blow the partial pressure changed from 55210-12

ndash 2110-10

and carbon mass

increased from 0754 ndash 299 wt A carbon mass of 787 had an oxygen partial pressure of

45110-13

whilst a lower carbon content of 153 wt had an oxygen partial pressure of

80610-11

The CO partial pressure however increased with increase in carbon composition in

the metal bath

When the oxygen flow rate increased a corresponding increase in the carbon removed (ΔC)

was observed For the first stage of the blowing process an increase in oxygen flow rate from

38867 ndash 6665Nm3 resulted in an increase in carbon removed from 506 ndash 728 wt The

second blowing stage had lower oxygen flow rates because of the carbon levels remaining in the

metal bath were around +- 2 wt In this stage oxygen flow rates increased from 125 ndash 28667

Nm3 and carbon removed (ΔC) from 016 ndash 2093 wt The slag showed that an increase in

basicity resulted in an increase in Cr2O3 in the slag As the basicity increased from 0478 ndash

1281 this resulted in an increase in Cr2O3 increase from 026 ndash 068 Nitrogen solubility in the

metal bath was investigated and it was observed that it increased with increasing Cr mass The

increase in nitrogen solubility with increasing Cr mass was independent of the nitrogen partial

pressures

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

ii

DEDICATION

I dedicate this work to my family they were patient with me whilst I dedicated more

time to my school work

iii

ACKNOWLEDGEMENTS

I would like to extend my thanks to the following people

1 My Supervisor Dr E Matinde whose guidance was the source of my strength even

when I thought I could not go on

2 Mr Jacques Beylefeld of XRAM Technologies who took his time to explain to me

some complicated thermodynamic principles

3 Mr Mark Verwayen a friend and brother who introduced me to AOD technology

iv

ABSTRACT

This study and the work done involves investigating the effects of different parameters on the

decarburization process of high carbon ferrochromium melts to produce medium carbon

ferrochrome and takes into account the manipulation of the different parameters and

thermodynamic models based on actual plant data Process plant data was collected from a

typical plant producing medium carbon ferrochrome alloys using AODs The molten alloy was

tapped from the EAF and charged into the AOD for decarburization using oxygen and nitrogen

gas mixtures The gases were blown into the converter through the bottom tuyeres Metal and

slag samples and temperature measurements were taken throughout the duration of each heat

The decarburization process was split into two main intervals namely first stage blow (where

carbon content in the metal bath is between 2-8 wt C) and second stage blow (carbon mass

below 2 wt ) The first and second blow stages were differentiated by the gas flow rates

whereby the first stage was signified by gas flow ratio of 21 (O2N2) whilst the stage blow had

11 ratio of oxygen and nitrogen respectively

The effect of Cr mass on carbon activity and how it relates to rate of decarburization was

investigated and the results indicated that an increase in Cr 6654 ndash 705 wt reduced carbon

activity in the metal bath from 0336 ndash 0511 for the first blowing stage For the second blowing

stage the increase in Cr mass of 6722 ndash 7165 wt resulted in an increase in C activity from

0336 ndash 057 The trend showed that an increase in chromium composition resulted in a decrease

in carbon activity and the same increase in Cr mass resulted in reduced carbon solubility

Based on the plant data it was observed that the rate of decarburization was time dependent that

is the longer the decarburization time interval the better the carbon removal from the metal

bath An interesting observation was that the change in carbon mass percent from the initial

composition to the final (ΔC) decreased from 1018 ndash 837 wt with the increase in CrC

ratio from 837 ndash 1018 This effect was attributed to the chromium affinity for carbon and the

fact that an increase in chromium content in the bath was seen to reduce activity of carbon It

was also observed that the effect of the CrC ratio was more significant in the first stage of the

blowing process compared to the second blowing stage A mass and energy balance model was

constructed for the process under study to predict composition of the metal bath at any time

interval under specified plant conditions and parameters The model was used to predict the

outcome of the process by manipulating certain parameters to achieve a set target By keeping

v

the gas flow rates blowing times gas ratios and initial metal bath temperature unchanged the

effect of initial temperature on decarburization in the converter was investigated The results

showed that the carbon end point with these parameters fixed decreased with increasing initial

temperature and this was supported by literature The partial pressure of oxygen was observed

to increase with decrease in C mass between the first and second blow stages For the second

stage blow the partial pressure changed from 55210-12

ndash 2110-10

and carbon mass

increased from 0754 ndash 299 wt A carbon mass of 787 had an oxygen partial pressure of

45110-13

whilst a lower carbon content of 153 wt had an oxygen partial pressure of

80610-11

The CO partial pressure however increased with increase in carbon composition in

the metal bath

When the oxygen flow rate increased a corresponding increase in the carbon removed (ΔC)

was observed For the first stage of the blowing process an increase in oxygen flow rate from

38867 ndash 6665Nm3 resulted in an increase in carbon removed from 506 ndash 728 wt The

second blowing stage had lower oxygen flow rates because of the carbon levels remaining in the

metal bath were around +- 2 wt In this stage oxygen flow rates increased from 125 ndash 28667

Nm3 and carbon removed (ΔC) from 016 ndash 2093 wt The slag showed that an increase in

basicity resulted in an increase in Cr2O3 in the slag As the basicity increased from 0478 ndash

1281 this resulted in an increase in Cr2O3 increase from 026 ndash 068 Nitrogen solubility in the

metal bath was investigated and it was observed that it increased with increasing Cr mass The

increase in nitrogen solubility with increasing Cr mass was independent of the nitrogen partial

pressures

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

iii

ACKNOWLEDGEMENTS

I would like to extend my thanks to the following people

1 My Supervisor Dr E Matinde whose guidance was the source of my strength even

when I thought I could not go on

2 Mr Jacques Beylefeld of XRAM Technologies who took his time to explain to me

some complicated thermodynamic principles

3 Mr Mark Verwayen a friend and brother who introduced me to AOD technology

iv

ABSTRACT

This study and the work done involves investigating the effects of different parameters on the

decarburization process of high carbon ferrochromium melts to produce medium carbon

ferrochrome and takes into account the manipulation of the different parameters and

thermodynamic models based on actual plant data Process plant data was collected from a

typical plant producing medium carbon ferrochrome alloys using AODs The molten alloy was

tapped from the EAF and charged into the AOD for decarburization using oxygen and nitrogen

gas mixtures The gases were blown into the converter through the bottom tuyeres Metal and

slag samples and temperature measurements were taken throughout the duration of each heat

The decarburization process was split into two main intervals namely first stage blow (where

carbon content in the metal bath is between 2-8 wt C) and second stage blow (carbon mass

below 2 wt ) The first and second blow stages were differentiated by the gas flow rates

whereby the first stage was signified by gas flow ratio of 21 (O2N2) whilst the stage blow had

11 ratio of oxygen and nitrogen respectively

The effect of Cr mass on carbon activity and how it relates to rate of decarburization was

investigated and the results indicated that an increase in Cr 6654 ndash 705 wt reduced carbon

activity in the metal bath from 0336 ndash 0511 for the first blowing stage For the second blowing

stage the increase in Cr mass of 6722 ndash 7165 wt resulted in an increase in C activity from

0336 ndash 057 The trend showed that an increase in chromium composition resulted in a decrease

in carbon activity and the same increase in Cr mass resulted in reduced carbon solubility

Based on the plant data it was observed that the rate of decarburization was time dependent that

is the longer the decarburization time interval the better the carbon removal from the metal

bath An interesting observation was that the change in carbon mass percent from the initial

composition to the final (ΔC) decreased from 1018 ndash 837 wt with the increase in CrC

ratio from 837 ndash 1018 This effect was attributed to the chromium affinity for carbon and the

fact that an increase in chromium content in the bath was seen to reduce activity of carbon It

was also observed that the effect of the CrC ratio was more significant in the first stage of the

blowing process compared to the second blowing stage A mass and energy balance model was

constructed for the process under study to predict composition of the metal bath at any time

interval under specified plant conditions and parameters The model was used to predict the

outcome of the process by manipulating certain parameters to achieve a set target By keeping

v

the gas flow rates blowing times gas ratios and initial metal bath temperature unchanged the

effect of initial temperature on decarburization in the converter was investigated The results

showed that the carbon end point with these parameters fixed decreased with increasing initial

temperature and this was supported by literature The partial pressure of oxygen was observed

to increase with decrease in C mass between the first and second blow stages For the second

stage blow the partial pressure changed from 55210-12

ndash 2110-10

and carbon mass

increased from 0754 ndash 299 wt A carbon mass of 787 had an oxygen partial pressure of

45110-13

whilst a lower carbon content of 153 wt had an oxygen partial pressure of

80610-11

The CO partial pressure however increased with increase in carbon composition in

the metal bath

When the oxygen flow rate increased a corresponding increase in the carbon removed (ΔC)

was observed For the first stage of the blowing process an increase in oxygen flow rate from

38867 ndash 6665Nm3 resulted in an increase in carbon removed from 506 ndash 728 wt The

second blowing stage had lower oxygen flow rates because of the carbon levels remaining in the

metal bath were around +- 2 wt In this stage oxygen flow rates increased from 125 ndash 28667

Nm3 and carbon removed (ΔC) from 016 ndash 2093 wt The slag showed that an increase in

basicity resulted in an increase in Cr2O3 in the slag As the basicity increased from 0478 ndash

1281 this resulted in an increase in Cr2O3 increase from 026 ndash 068 Nitrogen solubility in the

metal bath was investigated and it was observed that it increased with increasing Cr mass The

increase in nitrogen solubility with increasing Cr mass was independent of the nitrogen partial

pressures

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

iv

ABSTRACT

This study and the work done involves investigating the effects of different parameters on the

decarburization process of high carbon ferrochromium melts to produce medium carbon

ferrochrome and takes into account the manipulation of the different parameters and

thermodynamic models based on actual plant data Process plant data was collected from a

typical plant producing medium carbon ferrochrome alloys using AODs The molten alloy was

tapped from the EAF and charged into the AOD for decarburization using oxygen and nitrogen

gas mixtures The gases were blown into the converter through the bottom tuyeres Metal and

slag samples and temperature measurements were taken throughout the duration of each heat

The decarburization process was split into two main intervals namely first stage blow (where

carbon content in the metal bath is between 2-8 wt C) and second stage blow (carbon mass

below 2 wt ) The first and second blow stages were differentiated by the gas flow rates

whereby the first stage was signified by gas flow ratio of 21 (O2N2) whilst the stage blow had

11 ratio of oxygen and nitrogen respectively

The effect of Cr mass on carbon activity and how it relates to rate of decarburization was

investigated and the results indicated that an increase in Cr 6654 ndash 705 wt reduced carbon

activity in the metal bath from 0336 ndash 0511 for the first blowing stage For the second blowing

stage the increase in Cr mass of 6722 ndash 7165 wt resulted in an increase in C activity from

0336 ndash 057 The trend showed that an increase in chromium composition resulted in a decrease

in carbon activity and the same increase in Cr mass resulted in reduced carbon solubility

Based on the plant data it was observed that the rate of decarburization was time dependent that

is the longer the decarburization time interval the better the carbon removal from the metal

bath An interesting observation was that the change in carbon mass percent from the initial

composition to the final (ΔC) decreased from 1018 ndash 837 wt with the increase in CrC

ratio from 837 ndash 1018 This effect was attributed to the chromium affinity for carbon and the

fact that an increase in chromium content in the bath was seen to reduce activity of carbon It

was also observed that the effect of the CrC ratio was more significant in the first stage of the

blowing process compared to the second blowing stage A mass and energy balance model was

constructed for the process under study to predict composition of the metal bath at any time

interval under specified plant conditions and parameters The model was used to predict the

outcome of the process by manipulating certain parameters to achieve a set target By keeping

v

the gas flow rates blowing times gas ratios and initial metal bath temperature unchanged the

effect of initial temperature on decarburization in the converter was investigated The results

showed that the carbon end point with these parameters fixed decreased with increasing initial

temperature and this was supported by literature The partial pressure of oxygen was observed

to increase with decrease in C mass between the first and second blow stages For the second

stage blow the partial pressure changed from 55210-12

ndash 2110-10

and carbon mass

increased from 0754 ndash 299 wt A carbon mass of 787 had an oxygen partial pressure of

45110-13

whilst a lower carbon content of 153 wt had an oxygen partial pressure of

80610-11

The CO partial pressure however increased with increase in carbon composition in

the metal bath

When the oxygen flow rate increased a corresponding increase in the carbon removed (ΔC)

was observed For the first stage of the blowing process an increase in oxygen flow rate from

38867 ndash 6665Nm3 resulted in an increase in carbon removed from 506 ndash 728 wt The

second blowing stage had lower oxygen flow rates because of the carbon levels remaining in the

metal bath were around +- 2 wt In this stage oxygen flow rates increased from 125 ndash 28667

Nm3 and carbon removed (ΔC) from 016 ndash 2093 wt The slag showed that an increase in

basicity resulted in an increase in Cr2O3 in the slag As the basicity increased from 0478 ndash

1281 this resulted in an increase in Cr2O3 increase from 026 ndash 068 Nitrogen solubility in the

metal bath was investigated and it was observed that it increased with increasing Cr mass The

increase in nitrogen solubility with increasing Cr mass was independent of the nitrogen partial

pressures

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

v

the gas flow rates blowing times gas ratios and initial metal bath temperature unchanged the

effect of initial temperature on decarburization in the converter was investigated The results

showed that the carbon end point with these parameters fixed decreased with increasing initial

temperature and this was supported by literature The partial pressure of oxygen was observed

to increase with decrease in C mass between the first and second blow stages For the second

stage blow the partial pressure changed from 55210-12

ndash 2110-10

and carbon mass

increased from 0754 ndash 299 wt A carbon mass of 787 had an oxygen partial pressure of

45110-13

whilst a lower carbon content of 153 wt had an oxygen partial pressure of

80610-11

The CO partial pressure however increased with increase in carbon composition in

the metal bath

When the oxygen flow rate increased a corresponding increase in the carbon removed (ΔC)

was observed For the first stage of the blowing process an increase in oxygen flow rate from

38867 ndash 6665Nm3 resulted in an increase in carbon removed from 506 ndash 728 wt The

second blowing stage had lower oxygen flow rates because of the carbon levels remaining in the

metal bath were around +- 2 wt In this stage oxygen flow rates increased from 125 ndash 28667

Nm3 and carbon removed (ΔC) from 016 ndash 2093 wt The slag showed that an increase in

basicity resulted in an increase in Cr2O3 in the slag As the basicity increased from 0478 ndash

1281 this resulted in an increase in Cr2O3 increase from 026 ndash 068 Nitrogen solubility in the

metal bath was investigated and it was observed that it increased with increasing Cr mass The

increase in nitrogen solubility with increasing Cr mass was independent of the nitrogen partial

pressures

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

vi

TABLE OF CONTENTS

DECLARATION i

DEDICATION ii

ACKNOWLEDGEMENTS iii

ABSTRACT iv

ABBREVIATION NOTATION AND NOMANCLATURE xii

1 INTRODUCTION 1

2 LITERATURE REVIEW 6

21 Introduction 6

22 High Carbon Ferrochrome production 6

23 Production of Low and Medium Carbon Ferrochrome alloys 9

231 Unit processes in the production of low and medium carbon ferrochromium alloys 9

24 Refining process of High carbon Ferrochromium alloys 10

241 Refining of ferrochrome by chromium ore 11

242 Metallothermic Reduction 11

243 Converter Refining 13

25 Reactions amp thermodynamic considerations in the converter refining process 16

251 AOD decarburization process 18

252 Thermodynamic considerations for metal amp slag in ferrochrome refining 20

253 Activities of Cr and C in ferrochromium alloy refining 22

254 Interaction parameters in ferrochromium refining 24

255 Effect of blowing conditions on the extent decarburization 26

256 Effect of gas flow rates on decarburization 26

257 Initial Temperature of the liquid metal 28

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

vii

258 Rate equations in AOD refining 28

26 Energy balance in the AOD decarburization process 30

27 Summary 31

CHAPTER 3 32

3 METHODOLOGY 32

31 Introduction 32

32 Description of process under study 33

32 Plant data collection during the decarburization process 35

321 Blow periods data measurements 35

322 Data collation and modeling 36

33 Reduction stage after refining in the AOD 36

Summary 37

CHAPTER 4 38

4 Modelling Methodology 38

41 Introduction 38

42 Thermodynamic analysis of plant data 38

421 Oxygen partial pressure for oxidation of C Si and Cr 39

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si 40

43 Rate equations in the converter decarburization process 41

431 Rate equations at high carbon contents 42

432 Rate equations for low carbon levels 43

433 Equilibrium concentration of carbon 44

44 Mass and energy balance for the AOD process 45

441 Mass and energy balance construction procedure 46

441 Calculating the initial mass (moles) and energy of the metal bath in the converter 47

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200

viii

442 Calculating the mass (kmols) and energy balance for lime and dolomite 48

443 Kmols and energy calculations for FeSi in the reduction stage 50

444 Mass and energy balance calculations for AOD gases 51

45 Thermodynamics and Equilibrium considerations in AOD refining process 53

46 Interaction coefficients in metal and slag in the refining process 54

451 Activity coefficient calculations in metal phase 55

452 Activity coefficient calculations in the Slag Phase 56

CHAPTER 5 59

5 RESULTS AND DISCUSSION 59

51 Introduction 59

52 Carbon removal trend with decarburization for plant data 59

53 Mass balance for AOD refining stage 64

531 Decarburization process at high carbon content during the first stage of the blow 65

532 Comparison of the final chromium composition based on models and plant data 66

533 Effect of initial temperature on the extent of decarburization 68

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen 70

55 Effect of initial chromium content in the bath 72

56 Effect of O2 partial pressure on the extent of decarburization 72

57 Effect of CO partial pressure on the extent of decarburization 75

58 Fitting of model and plant data at lower carbon levels 76

581 Comparison of modelled and plant data in terms of carbon removal 76

59 Effect of gas flow rate on decarburization 77

510 Relationship between Chromium content and Cr activity in the metal bath 79

511 Effect of chromium on the activity of carbon in the alloy bath 80

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag 82

512 Nitrogen solubility in the alloy bath during decarburization 85

ix

513 Financial modelling of the AOD refining process 86

Summary 87

CONCLUSION 89

RECOMMENDATIONS 92

REFERENCES 95

APPENDICES 101

List of Figures

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom) 4

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995) 22

Figure 31 Process flow (adapted from the process under study) 34

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random

heats 60

Figure 52 Drop in carbon content with blowing interval 61

Figure 53 Change in carbon content (ΔC) with blowing time 62

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first

stage of blow 63

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage

of the blow 63

Figure 56 Comparison of model results vs plant data for the first stage of the blow 65

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon

content 67

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and

plant data during the second stage of the blow 67

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

69

Figure 510 Effect of initial temperature on end point carbon at low carbon contents 70

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for

Si Cr and C 71

x

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon 71

Figure 513 Effect of initial chromium on decarburization 72

Figure 514 Relationship between carbon mass vs PO2 73

Figure 515 Relationship between temperature and partial pressure at higher carbon levels 74

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon

levels 75

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in

the bath 76

Figure 518 Comparison of the decarburization between modelled and plant data 77

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing 79

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow 79

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath 80

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of

blowing) 81

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second

blowing stage) 81

Figure 524 Relationship between chromium oxide and activity in slag 83

Figure 525 Relationship between temperature and chrome oxide activity 84

Figure 526 Effect of slag basicity on activities of chrome oxides 84

Figure 527 Relationship between Chromium and nitrogen activity 86

List of Tables

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

7

Table 41 Analyses of Material inputs into the AOD 47

Table 42 Moles and Energy for metal 1600 oC 48

Table 43 Moles and Energy calculations for FeSi 50

Table 44 Moles and Energy calculations for Argon at 25 oC 52

Table 51 Summary of models used in the calculations of mass balance in the AOD 64

xi

List of Appendices

Appendix 1 AOD logsheet used in process under study 101

Appendix 2 Initial gas flows and gas ratios in model calculations 102

Appendix 3 Initial alloy bath analysis as charged into the AOD from the EAF 102

Appendix 4 Lime analysis as added into the AOD 102

Appendix 5 Dolomite analysis is added into the AOD 103

Appendix 6 FeSi analysis as added into the AOD 103

Appendix 7 Process input calculation summary 103

Appendix 8 Process output summary for mass and energy balance model 105

Appendix 9 Activity and activity coefficients calculations for mass and energy balance model

106

Appendix 10 first order interaction coefficient in liquid iron 106

Appendix 11 Calculated interaction coefficients for the energy and mass balance model 107

Appendix 12 Activity coefficients at infinite solutions 107

Appendix 13 Interaction energy between cations of major steel making slags 107

Appendix 14 Molar weights for oxides and elements in the energy and mass balance model 108

Appendix 15 Initial analysis of hot metal charged into the AOD for the process under study 109

Appendix 16 Second blowing stage initial data from the process under study 110

Appendix 17 Financial modeling of the AOD and EAF process 111

xii

ABBREVIATION NOTATION AND NOMANCLATURE

AOD Argon Oxygen Decarburizer

CLU Creusot Loire Uddeholm

EAF Electric Arc Furnace

CRK Ferrochrome Converter

SAF Submerged Arc Furnace

UTCAS Real time dynamic process control software

MCFeCr Medium carbon ferrochromium alloy

HCFeCr High carbon ferrochromium alloy

LCFeCr Low carbon ferrochromium alloy

DC Direct Current

VOD Vacuum Oxygen Decarburizer

HSC v7 H ndash enthalpy S ndash entropy C ndash heat capacity

1 INTRODUCTION

The production of medium and low carbon ferrochromium alloys is not a common practice

within the South African ferroalloys industry as most companies tend to focus on the production

of high carbon ferrochromium and charge chrome alloys Companies such as Samancor Chrome

have closed their IC3 plant which was used for the production of low and medium carbon

ferrochromium alloys due to unfavourable market costs and high electricity consumption

(MetalBulletin 2015)However the increased demand for low and medium carbon ferrochrome

alloys in the production of special alloys and stainless steel has resulted in the development of a

number of processes to reduce the content of carbon in the ferrochromium alloys

Processes such as the Simplex Duplex Perrin and converter decarburization were developed in

order to reduce the amount of carbon dissolved in ferrochromium alloys (Bhardwaj 2014) The

use of Argon Oxygen Decarburizers (AODs) and Creusot Loire Uddeholm (CLU) converters has

traditionally been used for stainless steel production (Pretorius and Nunnington 2002) but has

since found widespread applications in the production of low and medium ferrochromium alloys

(Kienel et al 1987)

High carbon ferrochromium and charge chrome alloys are produced from the reduction of

chrome lumpy ore or chrome concentrate from chrome fines concentration Submerged arc

furnaces in the production of high carbon ferrochromium and charge chrome alloys has become

the main technology used since it is economical (Gasik 2013) South Africa has over 75 of the

worldrsquos chromite reserves and with a huge reserve of coal (which is used to produce

metallurgical coke the main reductant in carbothermic reduction of chromite ores) thus making

South Africa one of the major producers of charge chrome (lt 55 wt Cr) in the world (Basson

et al 2007 Cramer et al 2004)

The present study is based on a South African company which utilizes electric arc furnaces

(EAFs) and Argon Oxygen Decarburization process (AODs) to produce medium carbon

ferrochromium alloy containing less than 10 wt carbon and +- 65 wt chromium in the

final alloy In this process high carbon ferrochrome fines are melted in an EAF and the alloy

melt is tapped into the AODs for converter refining and subsequently into casting ladles for

ingot casting or granulation Typically the initial carbon levels in the HCFeCr melts are around

4-6 wt and the target is to oxidize out the dissolved carbon to less than 10 wt Slag

2

formers are added in the form of burnt lime (CaO) and dolomite (MgOmiddotCaO) as well as

fluorspar to improve slag properties and to facilitate the dissolution of lime in the slag (Pretorius

and Nunnington 2002) Essentially the target basicity of the slag is at least 18

((CaO+MgO)SiO2) The current melting furnace utilizes refractory lining made of magnesia

based bricks depending on desired slag regime and cost From the EAFs the molten material is

tapped into transfer ladles before charging into manually operated AODs where the refining

process takes place

During the decarburization stage oxygen is blown into the AODs along with an inert gas (N2 or

Ar) which helps lower the partial pressure of the produced CO during oxidation of carbon After

the oxidation stage ferrosilicon is added in the reduction stage of the blowing process to recover

chromium that has been oxidized to slag (Pretorius and Nunnington 2002) In most cases the

pick-up of silicon in the metal does not exceed 05 wt When the desired liquid low or

medium ferrochromium metal composition is achieved the refined ferrochromium alloy is then

tapped into a teeming ladle and cast into ingots or is granulated

The AODs under study in this investigation are manually operated Initially the AODs were

automatically operated but the automatic operating model made the refining process take too

long and faced challenges in temperature control due huge discrepancies between the actual

measured temperature and model prediction In addition to these discrepancies operating costs

(particularly O2 N2 and refractory lining) of the process became too high thereby reducing the

profit margin Since the switch from automatic to manually operated AODs no technical studies

and research has been done to optimize the operation through the standardization of the critical

process parameters

As a result achieving process consistency in the operating parameters per heat such as slag

quality control decarburization time gas flow rates and ratios has never been achieved Based

on the aforementioned the main purpose of this research project is to conduct a parametric and

thermodynamic modeling of the AOD process based on the key process parameters link the

scientific models to actual process data and to optimize the AOD refining process based on the

results obtained from the thermodynamic models and parametric studies

3

SIGNIFICANCE OF THE STUDY

Ferroalloys are defined as alloys of iron and contain one more or elements such as chromium

manganese or silicon in significant quantities These alloys are critical to improving the

properties of steels For the purposes of this study the ferroalloy of concern involves the refining

of high carbon ferrochromium alloys The main element in this ferro alloy is chromium and is

mainly used in the stainless steel production Turkdogan and Fruehan 1999) Chromium is

responsible for increasing the corrosion resistance of stainless steel by forming a thin chrome

oxide layer on the surface which is stable and inert to chemical attack and reactions (Holappa

2002) Chromium as an alloying element is generally added in the form of high carbon

ferrochromium alloy (HCFeCr) but the carbon is an impurity especially in the production of low

carbon stainless steels and super alloys

With increasing demand for improved properties in steel impurities in steels have become a

major focus (Pande et al 2010) Typically the preference of lower carbon content in steel has

been a major driver in the production and consumption of low and medium carbon

ferrochromium alloys The lower carbon content enhances the steelrsquos mechanical properties such

as increased drawability and formability (AK Steel 2016) by reducing the precipitation of

chromium carbides Formation of these chromium carbides leads to intergranular corrosion

resulting in weld decays and failures at high temperatures due to steel sensitization which is

associated with high carbon contents in stainless steels (Bhonde et al 2007 Totten et al 2003)

In addition market prices of low and medium carbon FeCr alloys are comparatively higher than

for high carbon ferrochromium and charge chrome alloys For example the current market

prices for high carbon ferrochrome (7-9 wt C) range from USD$176-220kg whereas the

refined medium carbon ferrochromium alloys (08-10 wt C) at the time of the study was on

average USD$308kg (infominecommoditymine 2016) In essence higher market prices for

medium and low carbon ferrochromium alloys justify additional investments in the refining

processes Prices for commodities quoted in this section are prices at the time of the study and

are subject to fluctuations depending on market conditions (figure 11)

4

Figure 11 High carbon ferrochromium alloy prices for the past 7 years (infominecom)

The cost of producing lowmedium carbon ferrochromium alloys is driven by additional

infrastructure required for decarburization In particular the AOD converter costs are mainly

influenced by refractory materials consumption and the cost of gases In other operations

alternative gas options have been applied such as introduction of dry steam in order to reduce

converter costs when using argon as the inert gas (Beskow et al 2010) Nitrogen has also been

utilized as a substitute to try and bring costs down (Turkdogan and Fruehan 1999 Wei and Zhu

2002)

The process under study has experienced high AOD operational costs as no study was done on

the optimization of different process parameters thereby leading to poor costs control and alloy

recovery optimization of the process The findings of this study will improve the parametric

control of the decarburization process for production of lowmedium carbon ferrochromium

alloys using AODs This will aid in the reduction of operational costs which can be the basis of

the creation of cost models based on process consumables such as refractory linings oxygen

argon nitrogen among other cost drivers which will be tried for different rates of

decarburization

5

RESEARCH OBJECTIVES

The main objective of this study was to investigate the thermodynamic and parametric modeling

in the refining of high carbon ferrochrome alloys in manually operated AODs in the production

of medium carbon ferrochrome alloys The effect of key process parameters on the overall

process performance and final product quality was analyzed This in turn will lead in creation of

models to mimic the refining process to produce predictive changes in process parameters to

attain a specific required product specification In detail this project entails to

To determine optimum parameters for the refining process of high carbon ferrochromium

alloys to produce medium carbon ferrochromium alloys using manually operated AODs

To develop thermodynamic and mass and energy balance models (applicable) in the

converter refining process of high carbon ferrochromium alloys using manually operated

AODs

To apply the thermodynamic and mass and energy balance models to analyse the effects

of varying different process parameters on the refining process based on plant data

To develop a cost model for the production of medium based on key process variables

such as chromium losses to slag refractory materials consumption consumables usage

and other opportunity costs such as penalties and productivity variance

6

2 LITERATURE REVIEW

21 Introduction

The production of high carbon ferrochromium and charge chrome alloys through the reduction

of chrome ore by metallurgical coke or coal with fluxes has traditionally been done by use of

submerged arc furnaces (SAF) However carbon is an impurity in the production and use of

super alloys and some low carbon stainless steels thereby driving the need to produce lower

carbon content ferrochromium alloys In the past Metallothermic reduction techniques

(aluminothermic and silicothermic) have been able to produce ultra-low carbon with carbon

contents le 005 wt in the final alloys (Gasik 2013)

Other techniques of producing low carbon ferrochromium alloys include refining of FeCr by

chrome ore to reduce the carbon content through the reduction of Cr2O3 by carbon In addition to

this technique development of converter refining has contributed to the production of lower

carbon ferrochromium alloys from high carbon ferrochromium and charge chrome alloys In

essence use of converters include the use of argon oxygen decarburizers (AODs) ferrochrome

converters (CRK) and Creusot-Loire Uddeholm (CLU) coupled with the injection of an oxidant

and an inert gas into vessels through the tuyeres andor lance to remove the carbon in the molten

alloys (Heikkinen 2010)

Understanding of the different oxidation reactions and kinetics of these reactions in the AODs is

based on the knowledge from stainless steel production It is now possible to an extent to predict

the effect of changes in operational parameters such as gas flow rates during ferroalloys

converter refining (Wei amp Zhu 2002)

22 High Carbon Ferrochrome production

Ferrochromium alloys are generally produced in submerged arc furnaces (SAFs) from the

carbothermic reduction of chrome ores These alloys are a source of chromium which is a major

alloying element in the production of high value steel products Table 21 summarizes the

classification of high carbon ferrochromium and charge chrome alloys according to mainly the

chromium content in the alloys

7

Table 21 Classification of high carbon ferrochromium and charge chrome alloys (Gasik 2013)

Alloy Type Cr Si C P S

High Carbon Ferrochromium

60 - 70 lt2 lt8 lt0015 lt002

2 - 3 lt6 lt003 lt004

lt005 lt006

Charge Chrome

50 - 55 2 - 3 lt8 lt0015 lt002

3 - 4 lt003 lt004

4 - 5 lt005 lt006

As shown in Table 21 high carbon ferrochrome alloys typically contain carbon between 6-8 wt

C and chromium content in the range 60-70 wt Cr The production of such alloys requires

chromite ores that have higher CrFe ratio typically above 2 Charge chrome is produced from

chromite ores with lower CrFe ratio (usually between 15-16) and Cr levels of 50-55 wt and

sometimes even lower than 50 wt (Gasik 2013) Chromite ore typically exists in the form of

a spinel ((FeMg)O(CrFeAl)2O3) (Gasik 2013)

The carbothermic smelting route taken has an impact on the composition of particular elements

such as Si S and P The type and quality of reductant used will also impact on the final product

quality Usually reductants used are carbon based (coke coal charcoal anthracite etc) and

these are the main sources of S and P The quality of the ore will have an impact on the smelting

operations In particular the quality of raw materials can have significant impact on the

metallurgical efficiencies in the furnace (Gasik 2013)

The reaction for the reduction of chromite ore is shown in equation 21 (Ugwuegbu 2012)

11986211990321198651198901198744 + 4119862 = 2119862119903 + 119865119890 + 4119862119874 [21]

Generally submerged arc furnaces are used for smelting of Cr2O3 to Cr The reduction reactions

are promoted by use of reductants such as solid carbon in the form of either metallurgical coke

or coal or a combination of both Solid-gas reduction also occurs with CO and ore interaction as

illustrated by equations 22 and 23 (Chakraborty et al 2005)

8

11986211990321198743 + 3119862119874 = 2119862119903 + 31198621198742 [22]

119862 + 1198621198742 = 2119862119874 [23]

The chromium produced from chrome ore during carbothermic reduction will react with carbon

forming chromium carbides such as Cr3C2 Cr7C3 and Cr23C6 (Gasik 2013)

The DC arc furnace has a single preformed electrode with a hollow centre that is used to feed

charge into the furnace This electrode acts as the cathode and there is an anode at the bottom of

the furnace The DC furnaces use direct current unlike the SAF which utilizes alternating

current The DC furnaces are adaptable to the smelting of ore fines or concentrates (Hockaday et

al 2010)

In the submerged arc furnace chrome ore is charged into the furnace together with fluxes such

as limestone or quartz to adjust the slag basicity Most of the reduction of Cr2O3 occurs in the

coke bed which is located below the electrodes The initial slag that is formed in the furnace

contains significant amount of chromium which will be in the form of alloy entrained in the slag

or slightly altered chromite

The extent to which the spinel dissolves in the slag is also dependent on factors such as the

amount of slag produced the partial pressure of oxygen the length of the residence time the

material spends in the high temperature zone of the furnaces and slag composition Of

importance to note is also the chromium activities in the slag as this will impact on the

chromium recovery of chromium in the slag melt as this impacts on the financial viability of the

process and difficulties associated with discarding of chromium-containing slags as a waste

product (Gasik 2013 Holappa and Xiao 1992)

In the recent years technology has also been developed to smelt chrome fines and concentrates

by the use of plasma furnaces This development has been driven by the flexibility of plasma

furnaces to treat fines and easier and more rapid response to control compared to submerged arc

furnaces (Mac Rae 1992 Slatter et al 1986) The high carbon content in the high carbon

ferrochrome or charge chrome is a result of the carbon coming from the reductants in the form of

coke or coal which then exists as metal carbides of iron and chromium in solid ferrochrome

However the high carbon in the alloy has negative impacts on the alloy mechanical properties

9

In general some stainless steels and super alloys require ultra-low carbon contents In some

stainless steels higher carbon contents cause sensitization of the steels at elevated temperatures

as a result of the chromium carbides precipitating along grain boundaries As a result the

formation of carbides result in the areas adjacent to the grain boundaries becoming lsquostarvedrsquo of

Cr and less corrosion resistant leading to corrosion of the metal As a result there is need to

reduce the carbon content in high carbon ferrochromium and charge chrome alloys that are used

in the production of low carbon stainless steels (Bhonde 2007 Gasik 2013)

23 Production of Low and Medium Carbon Ferrochrome alloys

Low and medium carbon ferrochromium alloys have found widespread uses in the manufacture

of super alloys and super grades of stainless steels due to their lower carbon content Lower

carbon content in these alloys improves the chemical and mechanical properties of stainless

steels and other super alloys and this is principally achieved by using low carbon-containing

alloy additives such as low and medium carbon ferrochromium alloys (Heikkinen et al 2011)

Typically production of medium carbon ferrochromium alloy can be done through oxygen

blowing in a converter refining of HCFeCr or by the silico-thermic reduction of chromite ore

and concentrates Several techniques are used to lower the carbon content in ferrochrome alloy

and can be characterized as conventional processes such as metallothermic processes and non-

conventional methods that include processes such as decarburization of solid FeCr by vacuum

heat treatment process and use of an oxidant such as iron oxide or other oxidizing mixtures such

as steam and CO2 gas (Bhardwaj 2014)

231 Unit processes in the production of low and medium carbon ferrochromium alloys

The production of medium and low carbon ferrochrome alloys has been mainly driven by the

undesirable properties of carbon in ultra-low carbon stainless steels There has been a drive in

production of ultra-low carbon through the use of processes such as aluminothermic and

silicothermic reduction Converter technology has also revolved through use of oxygen steam

and in some cases use of CO2 gas Vacuum oxygen decarburization processes (VODs) have also

been used to produce ultra-low carbon ferrochrome (Wang et al 2015 Rick et al 2011)

The Perrin process is one of the processes used in the production of LCFeCr alloys and involves

the use of silico-chromiumferrosilicon as a reductant In the initial stages chrome ore

10

concentrate is blended with lime and charged into a kiln before charging into the furnace for

melting (Shoko et al 2004) The slag produced from the first stage of the process still contains a

quantifiable amount and contains in the range 24-26 wt Cr2O3 and is further reacted with

ferrosilicon chrome to recover the chrome units (Bhardwaj 2014)

The Duplex process is also another method which involves a silico-thermic reduction of a lime-

chromite slag to produce low carbon ferrochromium alloy (Ghose et al 1983) In this process

the ferrosiliconsilico-chromium is crushed to a fine size of 5-10mm and is then mixed with lime

and fed into a kiln before being charged into the electric furnace for melting (Ghose et al 1983)

An ore-lime melt produced in an electric arc furnace is tapped into a ladle and contains

approximately 40 wt CaO and between 25 ndash28 wt Cr2O3 (Ghose et al 1983)

Powdered (Ferrosilicon chromium alloy) FeSiCr is then added into the ladle to produce an alloy

with between 12-14 wt silicon content The amount of FeSiCr added depends on the analysis

of the FeSiCr the lime-ore melt and the desired product quality from this process The metal and

slag from the ladle are then poured and mixed thoroughly in a transfer ladle to facilitate the

reduction of Cr2O3 in the slag The transfer ladle is deslagged and the remaining metal is then

thrown back into the arc furnace with a new ore-lime melt batch to produce a final metal with le

1 wt Si (Bhardwaj 2014)

The third process used in the production of LCFeCr alloys is the Simplex process in which the

high carbon ferrochrome produced from a submerged arc furnace is crushed and milled in a ball

mill to produce a powdered material The milled powder is then mixed with Cr2O3 and Fe2O3

then briquetted before being decarburized by annealing at about 1350oC in a vacuum (Bhardwaj

2014)

24 Refining process of High carbon Ferrochromium alloys

As discussed in section 23 carbon has detrimental effects in some stainless steels and super

alloys as it reduces chemical and mechanical properties of these alloys The need to lower

carbon content has driven technology and innovations and these have taken into account the

costs associated with lowering of C Methods such as triplex process metallothermic reduction

converter and vacuum techniques were employed so that ultra-low carbon ferrochrome could be

directly produced (Bhonde et al 2007)

11

With the advent of VODs and AODs the chemical energy released from the oxidation of Si and

C in the high carbon ferrochromium alloys could be properly harnessed for fines re-melting

instead of reducing Cr2O3 in chrome ore resulting in reduced process costs for the refining

processes (Rick et al 2015) In detail some of the processes used in the production of low

carbon FeCr alloys are discussed in Sections 241 ndash 246

241 Refining of ferrochrome by chromium ore

The refining of HCFeCr alloys using Cr ore is done in a furnace based on the main assumption

that the chromium in the alloy exists as chromium carbides (Cr7C3) (Elyutin et al 1957) The

chromite is added into the (refining) furnace and can be represented by the equation 4

2

311986211990371198623 +

1

211986511989011986211990321198744 =

17

3119862119903 +

1

2119865119890 + 2119862119874 [24]

The viscosity of high chromium refining slag and high melting point (approx 2175K) makes this

whole process difficult This is due to the difficulty in separation between metal in slag due to

restriction to flow because of highly dense slag Slag fluidity is also a function of temperature

and a high slag melting point will result in a viscous slag at operating temperatures in the

furnace (Bhonde et al 2007) For decarburization to occur in this process high temperatures are

needed (Bhardwaj 2014)

242 Metallothermic Reduction

Metallothermic processes are usually exothermic in nature Metallic silicon (Si) and aluminum

(Al) are used in the silicothermic and aluminothermic processes respectively During the

metallothermic reduction he Al and Si are oxidized and this results in a sharp generation of heat

which is then utilized in the smelting process (Ghose et al 1983) In essence the aluminothermic

reduction is used to produce ultra-low carbon ferrochromium alloys while the silicothermic

reduction is used to produce low carbon ferrochromium alloys (Seetharaman et al 2014)

2421 Silico-thermic process

In the silicothermic process a mixture of chrome ore and lime is smelted in an electric arc

furnace (EAF) to produce a lime-ore melt The resulting lime-ore melt is then tapped into a ladle

where it is mixed with ferrosilicon chrome alloy and reaction is quite exothermic as silicon

being the reductant reduces the Cr2O3 in the melt to Cr This resulting slag produced as a result

12

of this reaction between ferrosilicon alloy and the lime-ore melt is a calcium silicate slag The

calcium silicate slag is then taken into a second ladle and mixed with molten high carbon

ferrochrome since the slag still contains chromium oxide that can be recovered The product of

this reaction is an intermediate product of ferrochrome silicon (Seetharaman et al 2014)

Low carbon ferrochrome can be produced using silico-chromeferrosilicon chrome as a reducing

agent An increase in silicon content will decrease carbon solubility in the metal (Bhardwaj

2014) The silico-thermic reduction process is capable of producing alloys with carbon content

below 003 wt C making it more preferable where very low carbon levels are desired and

such low carbon contents cannot be obtained using oxygen in the AOD process

The Perrin process is a typical example of the silicothermic reduction process The ferrosilicon

chromium alloy used in this process (35 ndash 43 wt Si) is produced in a smelting furnace such as

SAFs whilst chrome ore and lime are mixed in a different electric furnace to produce a lime-ore

melt contains between 24-26 wt Cr2O3 These two products (FeSiCr alloy and lime-ore melt)

produced from different furnaces are then intermittently mixed in a mixing ladle (mixing

commonly called lsquococktailingrsquo) The alloy produced in this mixing ladle has an average analysis

of 65-72 wt Cr and C of 003-005 wt In this process all reactions occur in the liquid

phase and reactions highly exothermic The main disadvantage of this process is the need to

synchronize all the processes to ensure no freezing of liquid material at any stage in the process

(Ghose et al 1983)

The Duplex process is relatively simpler than with the Perrin process as it uses ferrosilicon

chromium alloy in the solid form In this process ferrosilicon chromium alloy contains Si which

is the main reductant reducing Cr2O3 to Cr in the lime-ore melt The average analysis of FeSiCr

used is 38-40 wt Cr 43-45 wt Si and 003-005 wt C crushed to between 5-10mm in

size

The lime-ore melt from the EAF having an analysis of approx 25-28 wt Cr2O3 and 40 wt

CaO is mixed in a ladle with the ferrosilicon chromium alloy of 12-14 wt Si (which is

produced from a SAF) This mixture is then intermittently mixed in a transfer ladle in order to

recover as much residual Cr2O3 from the slag as possible and achieve a slag of lt1 wt Cr2O3

This slag is then discarded and the resultant intermediate alloy charged back into the furnace

with the lime-ore melt reducing silicon in the metal to lt1 wt The final slag usually contains

13

approx wt 3Cr2O3 which is generally higher than in the Perrin Process (Bhardwaj 2014

Ghose et al 1983)

2422 Alumino-thermic technique

The aluminothermic reduction process mainly applies for the in-situ production of ultra-low

ferrochromium alloy with at a smaller scale The slag produced from this process is refractory

(Al2O3 + Cr2O3) with a high melting point so lime (CaO) is added to the process in order to

lower the melting point of the slag thus also reducing the overall reaction temperature for the

process This slag is then kept fluid by addition of aluminum (Al) and NaNO3 into the charge

These additions will help in aiding metal recovery from slag The reactions below illustrate the

process occurring (Gasik 2013)

2

311986211990321198743 +

4

3119860119897 =

4

3119862119903 +

2

311986011989721198743 ∆119866119900 = minus309 + 0004119879 [25]

119870 = (11988611986011989721198743

23frasl

times 119886119862119903

43frasl

)(119886119860119897

43frasl

times 11988611986211990321198743

23frasl

)

Where

ɑi represents activities of Al2O3 Al Cr and Cr2O3

ΔGo is the standard Gibbs free energy

Aluminium as shown in equation 25 is used as the reductant of which the oxidation of Al is

exothermic and generates temperatures in the range 1800-2500oC In addition aluminothermic

reduction is promoted by lowering the Al2O3 activity in the generated slag This can be achieved

through addition of fluxes such as lime By preheating the feed to around 500oC the

aluminothermic reduction process is accelerated (Bhardwaj 2014)

243 Converter Refining

Converter technologies have widely been used in secondary metallurgy to refine melts that have

been produced in a primary process for example SAF or EAFs to reduce the carbon content in

the alloy (Heikkinen et al 2011) There are different types of converters currently being applied

namely the Creusot Loire Uddeholm (CLU) and Argon-Oxygen Decarburization (AOD)

The different converters basically use same principle (of blowing oxygen) for decarburization

Basically carbon is driven out of the bath through oxidation when oxygen is blown into the

14

alloy bath through the tuyeres andor top lance with an inert gas like Ar or N2 Steam has been

used in other instances as well as advances in using CO2 as an oxidant (Rick 2010)

2431 Utilization of gases in converter refining

The use of the AODs has dominated the stainless steel refining to achieve low carbon contents

since the introduction of the technology in the 1960s (Rick 2010) In the AOD oxygen is blown

with Ar or N2 as the inert gas The oxygen is the oxidant used for decarburization The oxidation

reactions are exothermic thereby raising bath temperature and as a result no external heat

source is used (Wei and Zhu 2002) The inert gases help lower the partial pressure of CO gas

and help drive carbon removal

The gases are introduced into the metal bath through the tuyeres at the bottom or a top lance

depending on converter design The gases also mix the bath making it turbulent therefore

increasing contact area between slag and metal and promoting slag-metal interface reactions

(Wei and Zhu 2002 Bhonde et al 2007) The molten metal charged into the AOD comes from

either an EAF (after scrap high carbon ferrochromium or charge chrome alloys melting) or

directly from the SAF as molten metal charged directly to the AOD (Wei and Zhu 2002)

Of importance in any AOD operation is the calculation and determination of production costs

These production costs include the cost of raw materials being charged into the converter the

tap-to-tap time which includes duration of each blowing operation and the life cycle and cost of

refractory material used to protect converter shell (Song et al 1992)

Due to the cost of Ar and problems with nitrogen pick-up into the metal processes such as the

CLU process were developed to lower costs of decarburization through use of a cheaper

alternative to Ar This process uses the fundamental background expressed in the equation [26]

This technology utilizes superheated steam which is mixed with oxygen compressed air and

nitrogenargon as bottom blown gases and the oxygen as a top blown gas (Beskow 2010)

1198672119874(119892) +2419119896119869

119898119900119897= 1198672(119892) + 121198742(119892) [26]

Rick (2010) stated that the use of a kg of steam equated to a substitution of 125nm3 of Ar or N2

0625nm3 of O2 and approximately 10kg of scrap The use of steam also helps in cooling the

15

bath thereby eliminating need for cooling scrap because the reduction of steam consumes heat

due to the endothermic nature of the reaction According to Rick (2010) nitrogen pick-up is also

eliminated through the use of steam and the reaction of dry steam in the converter

decarburization is depicted by equation 27

119862119903231198626 + 61198672119874 = 23119862119903 + 61198672 + 6119862119874 [27]

As shown in equation 27 H2 acts as the inert gas and O2 is the oxidant The use of steam will

also have an adverse effect of increasing hydrogen content in the metal which is undesirable

(Bhonde 2007)

In South Africa Columbus Steel in Middelburg implemented this method in 2008 (Beskow

2010) With the use of steam Columbus Steel managed to reduce their gas consumption costs by

reducing of the amount of crude argon consumed in the process by 20-25 thereby making it

possible for them to schedule production independent of argon availability (Beskow 2010) Due

to the endothermic nature of the reaction with steam it has the added advantage when there is

heat surplus in a converter by acting as an economic coolant compared to scrap addition during

the refining process (Bhonde 2007)

Use of carbon dioxide in the decarburization process has been investigated and applied to

decarburization as shown in the equation 28

119862119903231198626 + 61198621198742 = 23119862119903 + 12119862119874 [28]

The reaction above is endothermic and can only be thermodynamically feasible at high

temperatures and low CO partial pressure (Bhonde et al 2007) Wang et al (2015) concluded

that additions such as coolants were not necessary in a process where CO2 was used as the

oxidant as the endothermic decarburization reaction by CO2 consumed the excess energy

Furthermore bath temperature and temperature increases can be controlled by partially

introducing carbon monoxide to compensate for some oxygen in the AOD

In addition the use of CO as an oxidant improved chromium yield as a result of the weaker

oxidation ability of carbon monoxide thus reduced chromium oxidation Use of CO also leads to

16

improved refractory life span due to less thermal shock as is observed in the use of oxygen as the

excess energy from the oxidation reactions damages refractory lining (Wang et al 2015)

25 Reactions amp thermodynamic considerations in the converter refining process

In general converter refining takes lesser time than conventionally taken in a process like the

silico-thermic method thereby making it cheaper than most pyrometallurgical operations that

may require many furnaces The use of oxygen as the oxidant and the exothermic nature of the

oxidation reactions renders the process less costly as electric energy is only used for auxiliary

movement of the vessels Generally the main reactions occurring in converter refining

processes are the oxidation reactions of carbon chromium and silicon (Turkdogan amp Fruehan

1998)

As already discussed the CLU process uses dry steam oxygen argon compressed air and

nitrogen in the decarburization process The steam decomposes into hydrogen and oxygen at

elevated temperatures and since the endothermic decomposition reaction makes the temperature

control in the CLU efficient thereby eliminating need for adding scrap as a coolant As a result

this makes the CLU ideal for producing medium carbon ferrochrome and ferromanganese

especially in the later process where the evaporation of manganese from the bath due to high

temperatures is a serious health and environmental concern (Rick 2010 Bellow et al 2015)

The CRK converter is commonly used as an intermediate converter mainly for silicon and

carbon removal and for scrap melting while avoiding the oxidation of chromium The CRK

vessel design is similar to that of an AOD reactor but the main difference is mostly on the

purpose of use (Heikkinen 2012) In the CRK process oxygen is blown through the tuyeres or

via the top lance into the bath to reduce Si from initial 4-5 wt to below 05 wt The carbon

is also reduced from around 6-7 wt to 3 wt

The main reason for not decarburizing to carbon contents lower than 3 wt is to minimize the

oxidation of chromium oxidation which essentially tends to increase as the carbon content in the

bath decreases (Heikkinen 2012) From the work done by Heikkinen et al (2011) when the Si

content in the bath decrease to between 03-06 wt the carbon oxidation becomes the more

favourable reaction with oxygen until to below approx 25 wt thereafter the chromium

oxidation becomes significant in the CRK converter

17

Traditionally the AOD converter has been widely adopted in the stainless steel and mediumlow

carbon ferrochrome production (Wei and Zhu 2002) In the production of medium and low

carbon ferrochromium alloys the AOD converting process utilizes molten alloy from the EAF or

SAF in the form of high carbon ferrochrome andor charge chrome Depending on AOD vessel

design the oxygen is blown into the converter either from the bottom or from top lances The

key reactions in the AOD vessel involve the oxidation of C and Si and consequently Cr which

is then recovered in the reduction stage by addition of FeSi (Wei amp Zhu 2010) The typical

oxidation reactions that occur in the converter are shown in equations 29-212

[119862] + 1

21198742(119892) rarr 119862119874(119892) [29]

2[119862119903] + 3

21198742(119892) rarr (11986211990321198743) [210]

[119878119894] + 1198742(119892) rarr (1198781198941198742) [211]

[119865119890] + 1

21198742 rarr (119865119890119874) [212]

Other independent reactions also occur during the decarburization stage and these are shown in

equations 213-215 (Wei and Zhu 2002)

[119862] + (119865119890119874) rarr 119862119874 + [119865119890] [213]

∆119866119862 = ∆119866119862deg + 119877119879 ln ( 119875119862119874 (119886[119862] times 119886(119865119890119874))

2[119862119903] + 3(119865119890119874) rarr (11986211990321198743) + 3[119865119890] [214]

∆119866119888 = ∆119866deg119888 + ∆119866deg119862119903 + 119877119879119897119899 (11988611986211990321198743 (119886[119862119903]

2 times 119886(119865119890119874)3 ))

[119878119894] + 2(119865119890119874) rarr (1198781198941198742) + 2[119865119890] [215]

∆119866119878119894 = ∆119866119878119894deg + 119877119879 119897119899

119886(1198781198941198742)

119886[119878119894]119886(119865119890119874)2

The equations above show the Gibbs free energy changes ΔG which is a measure of the

thermodynamic driving force for the reactions to occur Thermodynamic principles state that if

ΔGo lt 0 at any particular temperature then the reaction is spontaneous and non-spontaneous if

the ΔGo gt 0 Therefore at process temperatures the oxidation reactions referred to in equations

29-215 are thermodynamically feasible (Wei and Zhu 2002)

18

251 AOD decarburization process

In general the AOD converting process uses oxygen for decarburizing (Wei and Zhu 2002) As

shown in equation [29] oxygen reacts with carbon to form carbon monoxide In the initial

stages of the blow the Oxygen Nitrogenargon ratio is usually high and the ratio is dependent

on the analysis of the liquid metal coming from the EAF In a typical AOD the oxygen and

nitrogenargon are blown into the bath through the tuyeres at the bottom of the vessel or a top

lance thereby stirring the bath as well (Wei and Zhu 2002)

According to Fruehan (1976) and Wei and Zhu (2002) the oxygen blown through the tuyeres

also oxidises Cr to form chromium oxide (Cr2O3) and the Cr2O3 in turn oxidises the carbon as it

rises through the metal bath due to the mixing effect of nitrogen or argon Fruehan (1976) further

assumed that the carbon oxidation is mainly determined by the rate of oxygen being blown for

higher carbon levels while for lower carbon levels the oxidation is mainly determined by the

rate of carbon mass transfer from the bulk of the melt to the bubble surface

Furthermore Fruehan stated that the carbon oxidation by Cr2O3 was mainly controlled and

dependent on the mass transfer of carbon to the surface of the inert gas bubbles resulting in

Cr2O3 being reduced to Cr However Wei and Zhu (2002) also clarified that the mass transfer of

carbon to the reaction sites was only rate determining at lower carbon contents whereas at

higher carbon composition the rate of oxygen blown into the metal bath controls the extent of

decarburization

2511 Thermodynamics of the decarburization reaction involving dissolved oxygen

The oxygen blown into the converter through the tuyeres andor the top lance reacts with carbon

as illustrated in equation [216]

[119862] + [119874] = 119862119874119892119886119904 119870 = 119875119862119874

119886119862times119886119874= 119890minus

∆119866119900

119877119879 [216]

Where

K is the equilibrium constant

PCO is the partial pressure of carbon monoxide

R is the molar gas constant

19

ɑi is the activity of carbon and oxygen

Where K is the equilibrium constant PCO partial pressure of CO R molar gas constant T is

temperature (in K) aC activity of carbon and ΔGo is the standard Gibbs free energy for the

reaction In order to drive the reaction forward lowering the CO partial pressure or increasing

oxygen activity will promote oxidation of carbon However an increase in partial pressure of

oxygen will also result in the oxidation of loss of Cr to slag as shown in equation 217

(Heikkinen et al 2011)

2[119862119903] + 3[119874] = (11986211990321198743) 119870 = 119886Cr2O3

1198861198621199032 times119886119874

3 = 119890minus120549119866119900

119877119879 [218]

The best option to reduce chromium oxidation is by lowering the partial pressure of CO in the

bubbles and this can be done through introduction of a diluent inert gas (Heikkinen et al 2011)

Nitrogen and argon gases are the main gases used as inert gases in the decarburization process

but nitrogen is preferred since is relatively cheaper than argon Heikkinen et al 2011 Wei and

Zhu 2002) However the use of nitrogen presents challenges such as the dissolution into alloy

will have a negative effect Nitrogen is known to cause strain ageing reduce ductility and

embrittlement of the heat affected zone (HAZ) for welded steels (Satyendra 2013)

In addition to lowering the partial pressure of CO is the AOD the use of inert gases such as N2

andor Ar also facilitates the attainment of higher equilibrium Cr percentages in the bath thereby

attaining lower carbon contents with minimum Cr loss to slag (Heikkinen et al 2011 Wei and

Zhu 2002) Inert gas blowing in the AOD has other advantages such as stirring of the bath to

ensure good contact area between the slag and metal interfaces (Heikkinen et al 2011) This is

achieved by reducing the oxygen to nitrogen andor argon ratio leading to more nitrogen or

argon and hence less oxygen in the bath resulting in reduced Cr oxidation (Heikkinen et al

2011)

2512 Reduction stage

When the desired carbon levels have been achieved in the decarburization stage the next stage is

to recover most if not all of the chromium that would have been oxidised to slag Essentially

the reduction is stage mainly involves the reduction of Cr2O3 to elemental Cr by using

20

ferrosilicon as a reductant (Heikkinen 2013) In this case the ferrosilicon reduces the metal

oxides as illustrated in in equation 219

3[119878119894] + 2(11986211990321198743) = 4[119862119903] + 3(1198781198941198742) [219]

During the reduction stage in the AOD the mixing of the metal bath is achieved by stirring with

an inert gas such as argon In the case where nitrogen has been used as a carrier gas during the

oxidation stage the introduction of argon gas also helps in the removal of any dissolved nitrogen

in the bath as argon has a higher partial pressure and a heavier molecule than nitrogen (Wei and

Zhu 2002) Addition of FeSi to the AOD is done at the reduction stage to reduce Cr2O3 from

slag to Cr thereby reducing losses to slag (Pretorius and Nunnington 2002)

252 Thermodynamic considerations for metal amp slag in ferrochrome refining

The metal and slag control in the AOD process is essential in order to get the desired slag and

metal quality As such it is important to understand the thermodynamic behaviour of metal and

slag in the AOD converter The slag chemistry and slag volume influence the bath temperatures

by insulating the bath against excessive heat losses protecting the refractory lining and also in

promoting the desulphurization of the metal bath (Pretorius and Nunnington 2002) The

desulphurization of the bath by slag-metal reactions takes place according to equation 18

(Pretorius and Nunnington 2002)

[119878]119887119886119905ℎ + (119862119886119874)119904119897119886119892 = (119862119886119878)119904119897119886119892 + [119874]119887119886119905ℎ [220]

AOD converting process utilizes basic slags target basicity 18 Essentially the typical slag

composition targeted for basicity calculations at the process under study is in the range CaO ge 40

wt MgO 10-12 wt and SiO2 le 30 wt Pretorius and Nunnington 2002) Basically the

use of basic slags and compounded by the reduced oxygen potential in the metal bath create the

ideal conditions for the desulphurization process (Pretorius and Nunnington 2002)

The slag basicity can be calculated as shown in equations 221-222 (Pretorius and Nunnington

2002)

119861 = 119862119886119874+119872119892119874

1198781198941198742 ge 18 [221]

119861 = 119862119886119874

1198781198941198742 ge 15 ndash 16 [222]

21

In general high slag volumes also have the negative effects on the efficiency of the AOD

process particularly in reducing the carbon removal negatively affects the desulphurization and

also the dissolution reactions in the reduction stage This in turn can be compounded by slag

carry over from the EAF or SAF increasing slag volume in the converter (Pretorius and

Nunnington 2002) Therefore the volume of slag in the refining process has to be minimized as

much as possible in order to enhance the escape of CO gas and also to reduce the wear of

refractory materials (Pretorius and Nunnington 2002 Xiao and Holappa 1992)

Burnt lime and dolomite are used as sources of CaO and MgO (basic oxides) needed for

increasing basicity The use of fluorspar as an additive is also practiced as it improves the lime

dissolution kinetics particularly in the reduction stage The addition of FeSi results in the

formation of CaSiO4 an inter-mediate phase due to reaction of SiO2 with the lime during the

reduction stage (Pretorius and Nunnington 2002)

The CaSiO4 produced further reacts with SiO2 to form CaSiO3 which then melts gradually to

form the final slag The use of CaF2 improves slag fluidity hence promote better mass transfer

rates and consequently improve desulphurization process as well as metal-slag separation

thereby reducing metal loss to slag in the converter upon tapping (Pretorius amp Nunnington

2002 Chuang et al 2002)

Figure 22 is a typical slag ternary diagram depicting the composition of stainless steel slags

(Pretorius and Nunnington 2002) Ideally a typical slag should have the analysis already

described in equations 221 and 222 The ternary diagram shown in Figure 22 gives an

indication of the liquidus temperatures for that particular analysis of slag desired

22

Figure 21 Slag ternary diagram for typical stainless steel slags (slag atlas 1995)

Poor and inconsistent slag control and monitoring in the AOD converter often results in high

losses of chromium to slag and this is particularly coupled with the negative impacts of high slag

viscosity which can lead to the high loss of metallics (Pretorius and Nunnington 2002 Ban-Ya

1993)

The use of fluxes in the smelting operations is done to modify the slag composition to ensure

that a slag with the required properties is produced Typical fluxes used in high carbon

ferrochromecharge chrome production are quartz dolomite or lime The quality of the slag

produced will have an impact as well on the process metallurgy melting temperatures as well as

the tapping conditions of the furnace (Gasik 2013)

253 Activities of Cr and C in ferrochromium alloy refining

The co-existence of the solute atoms such as Cr C and O in the bath has been extensively

studied in order to understand the bath refining process for Fe-C-Cr system (Ohtani 1956

Fruehan and Turkdogan 1999 Richardson and Dennis 1953) Richardson and Dennis (1953)

investigated the effect of Cr on the activity coefficient of carbon in a Fe-Cr-C bath by using

experiments which determined the conditions of equilibrium in the COCO2 mixture reaction

23

with C in the Fe-C-Cr melts The authors used a binary Fe-C system to determine the activity of

C in liquid Fe + C solutions and experimental results allowed for better accuracy in the

determination of carbon and iron activities

The C effect at a particular temperature on the activity coefficient of Cr can be depicted by the

interaction coefficient 120574119862119903119862 (Ohtani 1956) From these findings Ohtani (1956) proposed the

relationship between the interaction coefficient and the activity coefficient of Cr as

120574119862119903119862 = 120574119862119903120574119862119903

prime [223]

Where

120574119862119903 Is the activity coefficient of Cr in Fe-C-Cr alloys

120574119862119903prime is the activity coefficient of Cr in Fe-Cr

From the experimental work done by Ohtani (1956) it was shown that Fe-Cr alloys obey

Raoultrsquos law and that the alloys tend to show a negative deviation from Raoultrsquos law when C is

added as a third element to Fe-Cr binary alloys Essentially Raoultrsquos law states the partial

pressure of each component in an ideal liquid mixture can be obtained from the product of that

specific component with its composition (mole fraction) in that particular mixture in question

(Gaskell 2013)

In addition to the influence of C on Cr the effect of Cr on C was also investigated and the

relationship is represented as shown in equation 224 (Ohtani 1956)

120574119862119862119903 = 120574119862120574119862

prime [224]

Based on the studies by Richardson and Dennis (1953) to determine the equilibrium for COCO2

mixtures with C in Fe-C-Cr alloys it is possible to evaluate effect of Cr on the activity

coefficient of carbon in the bath The results obtained from this by the authors work made it

possible for calculation of thermodynamic properties for Fe-C with better accuracy Ohtani

(1956) proposed that chromium has affinity for carbon and was shown by the change in

solubility of graphite on addition of chromium in Fe-Cr-C alloys This was attributed to the drop

in carbon activity in iron solutions upon addition of chromium

24

According to Ohtani (1956) the interaction energy between Cr and C is higher than that of Fe-C

atoms and hence the distribution of C in relation to Cr atoms is not random Ohtani (1956)

further proposed that carbon atoms in molten Fe-Cr-C alloys tend to preferably agglomerate

around chromium atoms than iron atoms thus exist with a higher proportion of Cr atoms around

them Therefore the affinity between the Cr and C atoms makes it difficult for the

decarburization to occur in the converter especially at lower C contents (Ohtani 1956) In other

words the Cr content in the metal bath will affect the extent to which C can be oxidized without

the co-oxidation Cr (Ohtani 1956)

254 Interaction parameters in ferrochromium refining

Interaction parameters are defined as a measure of interactions that occur between atoms or

molecules in a solution This can be the interaction between these moleculesatoms with one

another or with the medium or solvent in which they are dissolved (Tados 2013 Ohtani 1956)

From the work done by Fuwa and Chipman (1959) on the activity of carbon in Fe-C-X systems

reveal the researchers concluded that interaction parameters are linked to sites of the elements

(X) on the periodic table and tend to vary with the atomic number of the element X

Wada et al (1951) stated that the interaction parameters were related to the excess free energy of

mixing that is the interaction parameters depended on interaction energies between solute and

solvent atoms (or solute and solute atoms) Wagner (1952) developed the first order free energy

interaction coefficients for dilute multi-component solutions which is the coefficient in the

Taylor series expansion for the logarithm of the activity coefficients and represented as shown in

equations 225 and 226

120576119894(119895)

equiv [120597119897119899120574119894

120597119883119895]1198831rarr1 [225]

119897119899120574119894 = 119897119899120574119894119900 + sum 120576119894

(119895)119883119895

119898119895=2 + ℎ119894119892ℎ119890119903 119900119903119889119890119903 119905119890119903119898119904 [226]

Bale and Pelton (1989) developed modifications to the unified interaction parameter formalism

The formalism proposed by Pelton and Bale (1989) reduces to the formalism developed by

Wagner (1952) at infinite dilution (by neglecting all other higher orders except for the first order

terms) to an expression that is linear with respect to molar fractions of solutes in dilute alloy as

shown in equation 226

25

The advantage with the unified interaction parameter formalism over the standard formalism is

that it remains thermodynamically consistent at finite concentrations whereas the latter does not

It should be noted however that at infinite dilutions the unified interaction parameter formalism

reduces to standard formalism and keeps the numerical values of parameters as well as the

notation (Bale and Pelton 1966) This calculation for interaction parameters and activity

coefficients was used in the process under study to calculate activities and activity coefficients

for Cr C Si for the study and mass and energy balance model development

Taking a system with (N+1) components where there is N solutes and a solvent and limiting it

to first order interaction parameters the unified interaction parameter formalism can be

expressed as shown in equation 226 (Bale and Pelton 1966)

119897119899120574119894 = 119897119899120574119894119900 + 119897119899120574119878119900119897119907119890119899119905 + 12057611989411198831 + 12057611989421198832 + 12057611989431198833 + ⋯ + 120576119894119873119883119873 [226]

In additionfurthermore the equation for the activity coefficient of the solvent is expressed as

119897119899120574119904119900119897119907119890119899119905 = minus1

2sum sum 120576119895119896119883119895119883119896

119873119896=1

119873119895=1 [227]

Where

Xi is the mole fraction of a component i

εij interaction coefficients

γi activity coefficient of component i

γsolvent activity coefficient of solvent

For first order interaction parameters in ternary systems with a solvent and two solutes namely

solute 1 and solute 2 the following quadratic formalisms were derived (Pelton and Bale 1989)

119897119899120574119904119900119897119907119890119899119905 = minus (12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832 [228]

1198971198991205741 = 1198971198991205741119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576111198831 + 120576211198832 [229]

1198971198991205742 = 1198971198991205742119900 + [minus (

12057611

2) 1198831

2 minus (12057622

2) 1198832

2 minus 1205761211988311198832] + 120576221198832 + 120576121198831 [230]

26

The above equations are valid for all compositions and are analogous to the quadratic formalism

proposed by Darken (1967) which is thermodynamically consistent at finite concentrations

With the determination of these models for activities and activity coefficients the same models

were used to determine the interaction between interaction of Cr Si and C with Fe as the solvent

(and the oxides of these elements) in the refining of high carbon ferrochromium alloy in the bath

in order to predict behaviour of these elements and oxides in the converter refining

255 Effect of blowing conditions on the extent decarburization

In the refining stage the carbon content in the bath gets depleted as decarburization proceeds

until critical carbon level is attainted Turkdogan and Fruehan (1999) defined the critical carbon

as that carbon content below which chromium oxidation commences These authors further

stated that critical carbon level increased with the chromium content (or activity of chromium) in

the alloy bath However critical carbon in the refining process is determined by a number of

factors such as thermodynamic and kinetics considerations temperature and chemical

composition of the alloy bath blowing rates of argon and oxygen (or any other inert gas used)

the configuration and geometry of the vessel activity coefficients of the individual elements in

the alloy bath and the mixing and stirring effect of the blown gases (Wei and Zhu 2002)

According to Wei and Zhu (2002) the blowing time is a function of gas flow rates and a

balance between ratios of oxygen to nitrogen or argon (Wei and Zhu 2002) When carbon

reaches a critical content Cr loss to slag increases with further blowing time and the Cr loss to

slag can be represented as shown below in equation 231

∆[119862119903] = 4119872119862119903

311988210minus21198731198742

119905 minus 10minus2119882

2119872119888∆[119862] [231]

Where MCr Mc are molecular weights for Cr and C respectively W weight of steel 1198731198742 molar

flow rate of O2 and Δ[C] change in carbon content (Turkdogan and Fruehan 1999)

256 Effect of gas flow rates on decarburization

Volumetric gas flow rates have an influence on oxidation reactions that occur within a converter

The O2Ar gas ratios play a major role in the efficiency of the decarburization process (Wei et al

2010) A reduction in O2 Ar ratio with the blowing intervals improves the effectiveness of the

27

refining process Fruehan (1976) stated that the oxygen volumetric flow rate was critical for

decarburization at higher carbon levels in the alloy bath Wei and Zhu (2002) went on further to

state that the gas flow rate has a significant effect on decarburization and the oxidation of

elements in the alloy bath At critical carbon the reduction of oxygen gas flow rate and increase

in argon flow rate dilutes the carbon monoxide produced from carbon oxidation thereby

promoting further carbon oxidation at lower carbon contents in the bath (Wei and Zhu 2002)

As decarburization proceeds the C remaining in the bath becomes rate determining as an

increase in nitrogen and decrease in oxygen flow rates will promote decrease in partial pressure

of CO and thereby promoting further decarburization (Wei et al 2010) The inconsistent control

of gas flow rate translates to higher operational costs due to un-optimized gas consumption and

blow times The inconsistent blowing cycles will increase costs on gases reduced refractory

lifespan Cr loss in slag and overally decreased productivity It is quite prudent to have a

standardized blowing rate with consistent gas flows to improve on process efficiencies

The decarburization rate increases as the blowing process proceeds and this results in the

increase of the bath temperature from the exothermic oxidation reactions During this stage the

decarburization rate is directly influenced by the volumetric flow rate of the oxygen gas As the

carbon level decreases it reaches a critical point value (critical carbon level) where the

decarburization rate then becomes more dependent on mass transfer of the carbon in the bath

(Wei et al 2010)

According to Turkdogan and Fruehan (1998) the critical carbon content during refining is that

carbon content below which Cr in the bath starts getting oxidized The critical carbon will

increase with decrease in bath temperature and increase in Cr content or Cr activity (Turkdogan

and Fruehan 1999) When oxygen is blown into the converter Cr gets oxidized first to Cr2O3

which is further reduced by the solid carbon as it rises into the slag-metal emulsions (Turkdogan

and Fruehan 1999)

11986211990321198743 + 3119862 = 2[119862119903] + 3119862119874 [232]

119889119862

119889119905 = minus

120588

119882sum 119898119894119894 119860119894[119862 minus 119862119894

119890] [233]

28

Where ρ is the steel density W is weight of the metal mi is mass transfer coefficients Ai the

surface areas Cei equilibrium carbon content and subscript i individual reaction sites eg top

slag or rising bubble

257 Initial Temperature of the liquid metal

In most cases crude steelFeCr from EAFSAF is conveyed to the AOD in a transfer ladle as

liquid alloy The initial temperature of the metal charged into the converter is important In other

words a lower carbon content and lower Cr loss in slag is achievable at the end of the refining

stage with a higher starting temperature of the liquid metal This will impact positively on the

amount of FeSi needed for recovery of Cr from slag (Wei and Zhu 2002 Fruehan 1976

Turkdogan and Fruehan 1999) Due to the manual process implemented in this process data the

impact of initial temperature on decarburization has not been observed

Wei and Zhu (2011) studied the impact of initial temperature on decarburization All other

process parameters were kept constant and the initial temperature of the bath was varied by +-

10K It was observed that an increase by 10K resulted in a decrease in the final end point carbon

and a decrease of initial temperature by -10K resulted in an increase in end point carbon content

It was however noted in that study that consistent implementation of this study on process level

was difficult as every heat characteristic differs from the next heat in process operation

258 Rate equations in AOD refining

The rate of decarburization can be described by rate equations which can be used as predictive

models in the refining process At higher carbon levels in the alloy bath the oxygen flow rate

determines carbon oxidation and carbon liquid mass transfer to reaction interface is rate limiting

at lower carbon contents (Wei and Zhu 2002 Fruehan 1976) By considering these conditions

of carbon oxidation for higher and lower carbon levels it is possible to model rate equations for

oxidation of elements in the alloy bath (Wei and Zhu 2002) Wei and Zhu (2002) and Fruehan

(1976) defined lower carbon contents as that carbon content below the critical carbon content

where carbon mass transfer becomes rate limiting and not the oxygen flow rate into the alloy

bath

29

2581 Rate Equations at higher carbon levels

The rate of decarburization for the refining process at higher carbon levels in the metal bath is

mainly dependent on the rate of oxygen blown into the process (Wei and Zhu 2002 Fruehan

1976) The average losses of the main elements in the bath like silicon carbon and chromium are

described in the equations 234 ndash 236 below (Wei and Zhu 2002 Fruehan 1976 Diaz et al

1997)

119889[119862]

119889119905= minus

2120578119876119874

22400times 119909119862 times

100lowast119872119862

119882119898 [234]

119889[119862119903]

119889119905= minus

2120578119876119874

22400times 119909119862119903 times

100lowast119872119862119903

15lowast119882119898 [235]

119889[119878119894]

119889119905= minus

2120578119876119874

22400lowast 119909119878119894 lowast

100lowast119872119878119894

2lowast119882119898 [236]

where

119876119874 is flow rate of oxygen

Mi is the molar mass of the substance i

X is the molar fraction of each element i

Wm is the mass of the alloy bath charged into the converter

120578 is the utilisation ratio of oxygen

2582 Rate Equations for the low carbon contents

The authors Wei and Zhu (2002) and Fruehan (1976) defined lower carbon contents in the alloy

bath to be lower than the critical carbon of the alloy bath under which carbon mass transfer to

the reaction interface becomes rate limiting and not oxygen flow rate into the alloy bath At low

carbon concentrations the main oxidation reactions are that of chromium and carbon and these

reactions increase the alloy bath temperature as these reactions are exothermic in nature The

equation that appropriately describes this scenario is equation 237 (Wei amp Zhu 2002)

minus119882119898119889[119862]

119889119905= 119860119903119890119886120588119898119896119888([119862] minus [119862]119890) [237]

Where

Wm - mass of liquid alloy (g)

[C] - mass of concentrate of carbon in molten alloy (wt )

Area ndash total reaction interface (cm2)

30

120588119898 ndash density of alloy (gm3)

kc ndash mass transfer of carbon in molten alloy (cms)

[C]e ndash equilibrium concentration of carbon in molten alloy at reaction interface (wt )

26 Energy balance in the AOD decarburization process

It is important to look at the energy balance and requirements for decarburization The energy in

the converter comes from oxidation reactions of carbon chromium and silicon as illustrated in

equations in equations 29 ndash 215 (Heikkinen et al 2011) These oxidation reactions release heat

as the elements in the metal bath get oxidised (Wei and Zhu 2002) The molten alloy charged

into the converter along with all the gases (argonnitrogen oxygen and the exhaust gases) all

carry heat in the converter Slag formers added in the form of lime and dolomite also take in heat

in the formation of slag (Wei and Zhu 2002) Wei and Zhu (2002) went on to state that as heat

rises in the converter due to the oxidation reactions of the elements (of which the reactions are

exothermic) heat losses through exhaust gases radiation when converter is tilted to take samples

and check bath temperature and by conduction and adsorption through the refractory lining are

experienced Pretorius and Nunnington (2002) also stated that large amounts of slag formers

could lead to high volumes of slag and insulate the bath hence affecting high refractory wear and

inhibit carbon removal efficiency

A general expression of energy balance (Wei amp Zhu 2002) during decarburization is as shown

in Equation 238

∆119867119862 + 120549119867119878119894 + 120549119867119872 + 119864 = 119862119901prime ∆119879 + 119902119900119905 [238]

∆119919119914 is the heat of oxidation of Carbon

ΔHSi is the heat of oxidation of Silicon

ΔHM is the heat of oxidation of Chromium and Iron

E is the electrical energy

Crsquop is apparent heat capacity of the metalrefractory system

∆T is the temperature change during the decarburization period

qo represents steady state external heat loss rate

t is the total elapsed time from start of oxygen blowing until the start of the reduction stage

31

Some of the bath temperature is partly lost due to conduction and adsorption by the refractory

lining and shell during the refining process When the converter is tilted to facilitate sampling

and temperature checks radiation losses are also experienced Additions of coolants for

temperature control also lower bath temperature (Wei amp Zhu 2002)

27 Summary

High carbon ferrochromium and charge chrome alloy production is mainly through the

carbothermic process (with use of SAF) where carbon is used as the main reductant and fluxes

such as limestone and quartz are used to achieve required slag chemistry and properties (Gasik

2013) Technology advancement has resulted in the development of plasma furnaces in the

manufacture of high carbon ferrochromium and charge chrome alloys through the smelting of

chromite fines However the use of high carbon ferrochromium and charge chrome in the

production of stainless steel has led to the development of technology for carbon removal as

carbon is an impurity in super alloys and low carbon stainless steels production

Different technologies such as the metallothermic processes and ferrochromium alloy refining

with chrome ore have been developed to lower carbon content However the development of

converter refining has found greater use as it is cheaper Of particular interest in the converter

technology is the use of converter technology mainly in the production of stainless steel

However this technology has been harnessed in the production of medium carbon

ferrochromium alloy to reduce carbon content in the alloy The process under study has

harnessed the use of AOD technology in the production of medium carbon ferrochromium and

forms the basis of this investigation

32

CHAPTER 3

3 METHODOLOGY

31 Introduction

The parametric study in the decarburization process of high carbon ferrochromium alloys in the

AOD converter enhances the understanding and process control of the thermodynamics involved

in the process The carbon in the liquid alloy is oxidised to CO gas which is then driven out of

the bath thereby lowering the carbon in the metal (Wei ampZhu 2002) Chromium is also oxidised

particularly around the tuyere area and then as the Cr2O3 rises with the argonnitrogen bubbles it

oxidizes the carbon thus getting reduced to chromium (Fruehan 1976) Fruehan (1976) also

assumed that at lower carbon levels the liquid mass transfer of carbon to the bubble surface

determines the carbon oxidation by Cr2O3 but at higher carbon levels it is primarily determined

by the amount and rate of oxygen flow blown into the vessel

In this study the process under study was investigated to evaluate process performance for

manually operated AODs in the manufacture of medium carbon ferrochromium alloy The AOD

processing route has been traditionally used for stainless steel production and most research

done has been to understand thermodynamics and kinetics for stainless steel production (Wei

and Zhu 2002 Freuhan 1956) The automated operation was abandoned after the simulation

prediction for temperature and carbon levels deviated significantly from the results obtained

from sampling and manual temperature measurements rendering process control difficult The

UTCAS system (real-time dynamic process control software) was used to run the AOD on

automatic but was since stopped though data like pressure and volumetric flow can still be

obtained on the display

Actual gas flows and gas pressures were displayed on the screen from the UTCAS display and

these values were then used to calculate normal gas flow rates for the plant data collected The

use of nitrogen as the inert gas of choice was chosen for the process under study as the cost of

argon gas was higher than that of nitrogen Since the onset of the manual operation no scientific

study has been undertaken to improve process performance Plant data was collected for 30

AOD heats in order to understand the decarburization rates and effect of different parameters on

the process over a 30 day interval The limited time frame for data collection was a result of time

33

constraints for compiling of the final thesis thus extended data collection interval was not

possible for the plant under study This process data was collected by the control room operator

that is filled onto an AOD log-sheet (template shown in the Appendix 1)

32 Description of process under study

The process under study involves the utilization of EAFs to re-melt high carbon ferrochromium

alloy fines (gt 10mm in size) with analysis of the metallic fines ranging from 63 ndash 68 wt Cr

07 ndash 15 wt Si and carbon gt 7 wt The process flow is shown in figure 4 The high carbon

ferrochromium alloy fines are charged into the electric furnaces mixed with a blend of slag

formers (burnt lime and dolomite) which form the basis for slag formation The melting process

in the furnace on average has duration of 3hrs When the melting is complete the electric

furnaces are tilted to tap the slag into the slag pots

Upon attaining a temperature of approximately 1700oC after the last deslagging the molten alloy

is then tapped into a transfer ladle and the net weight of the alloy measured Care is taken to

ensure that slag carry over to the AOD is avoided This is to ensure that the slag volume in the

converter is controlled A high slag volume in the converter has been observed to reduce rate of

decarburization and high temperatures in the bath (gt 1760oC) resulting in reduced

oxygennitrogen gas ratios to try and lower the bath temperature (Wei and Zhu 2002 Pretorius

and Nunnington 2002) The high slag volume insulates the bath resulting in very high bath

temperatures being experienced (Pretorius and Nunnington 2002)

The tapped molten alloy is charged into the AOD when the transfer ladle is transported using an

overhead crane The AOD is titled to allow for transfer of molten alloy to the vessel At this

moment a quick immersion thermocouple (QIT) is used to measure the initial temperature of the

bath The converter is then tilted back to the vertical position where the first stage of blowing

takes place The first stage of the blowing process involves decarburizing a carbon content gt2

wt This stage also involves blowing a gas flow ratio of 31 oxygen and nitrogen as at this

stage carbon is sufficiently high and the rate of oxygen blown into the AOD by the two bottom

tuyeres is rate determining (Fruehan 1976)

When the measured carbon level is less than 2 wt this signifies the start of the second stage

of blowing In this stage the carbon is considered low enough to adjust the oxygennitrogen ratio

34

to 11 respectively The carbon content is sufficiently low for carbon content liquid mass transfer

to the nitrogen bubbles as the Cr2O3 rises in the bath to be rate determining This means the rate

of oxygen flow into the converter is no longer determining the decarburization process (Wei and

Zhu 2002 Fruehan 2002)

Upon attaining a carbon content of gt 1 wt carbon the decarburization process moves into the

reduction phase where ferrosilicon alloy (FeSi) is added in order to recover the chromium that

has been oxidized to slag In this stage of the process silicon in the FeSi reduces Cr2O3 to Cr and

this chromium reports to the metal bath To reduce loss of metallic elements to slag fluorspar is

added to improve slag fluidity and also reduce alloy entrainment in slag (Pretorius and

Nunnington 2002) Argon is then blown into the metal bath and oxygen and nitrogen

discontinued The argon is meant to stir the bath during reduction stage (Wei and Zhu 2002)

The metal is then tapped into molded pans and sent for crushing once it has sufficiently cooled

to be crushed

Figure 31 Process flow (adapted from the process under study)

For the duration of the time interval under study same operating conditions (such as the quality

of the high carbon fines melted) remained the same thus eliminating influence of change in

process conditions which can change process performance

35

32 Plant data collection during the decarburization process

The aim of collecting this data was to study the process parameters and their influence on the

decarburization process The molten metal tapped from the EAF was conveyed to the AOD

using a transfer ladle hooked to an overhead crane and the mass of the molten alloy was

determined and recorded onto the log sheet This formed the basis of calculating the initial

molten metal weight charged into the AOD

Samples of the molten alloy were taken from the EAF during tapping stage and then sent to the

laboratory for chemical compositional analysis The analyses of major elements C Si Cr and

Fe were conducted using a Spectromaxx Metal Analyzer and the results were then taken to the

EAF and AOD control rooms and recorded on the log sheets The initial temperature of the melt

charged into the converter was also measured using dipQIT thermocouples Slag samples were

also taken for analysis on the X-ray Fluorescence Analyzer

Once the molten metal has been transferred to the AOD the decarburization process

commences and during the first and second blowing stages the metal melt samples were also

taken in order to check the extent of decarburization and to optimize the process gas flow rates

gas pressure and bath temperatures for optimal process control These process parameters were

recorded and gas flow rates and system pressure measurements based on the UTCAS system

(computer control system designed for converter refining)

The exact masses of slag formers and FeSi added were obtained from the scales at the AOD

batching plant The final weight analysis and temperatures were also measured after the final

specifications were achieved on Carbon and Silicon

321 Blow periods data measurements

First stage of the blowing process involved blowing oxygen and nitrogen at higher carbon levels

in the alloy bath and was done for at least 20 mins after addition of slag formers and initial

temperature measurement

The initial temperature measured at the AOD was a function of the temperature at which the

same tap was made at the EAF and the time it took to transfer the same tap from the EAF to the

AOD This would influence temperature pick up in the AOD from the start of the blowing

36

process as observed hence the lower the initial temperature the longer it took for temperature to

pick up to between 1720 ndash 1750oC If temperature was less than 1720

oC blow 1 continued until

desired temperature was reached However the change from first to the second blowing stage is

done once carbon goes down to le 20 wt The second blowing stage involved taking down

carbon from less than 2 wt obtained from blow 1 to around 08 ndash 10 wt This was usually

between 15 ndash 20 mins depending on the carbon level

The most significant change in the second stage of blowing was the reduction in Oxygen

Nitrogen ratios The same measurements as in blow 1 were repeated until the end of blow 2 If

carbon was analyzed to be between 10 ndash 12 wt then the blow interval was further increased

by another 7-10 mins If however if it still remained above 12 wt after the second blow then

the blow time was increased by 15-20 mins After the extension of blow 2 to accommodate for

higher carbon levels the carbon was observed to go under 10 wt and the refining stage

ensued

322 Data collation and modeling

A mass and energy balance model was constructed based on the collected plant data The data

obtained from the process log sheets for the 30 heats was interpreted in terms of mass and

energy balance parameters such as interaction parameters energy contribution of each

elementoxide and Gibbs free energies amongst other parameters

Once the energy and mass balance model was constructed the different parameters discussed in

the literature review were manipulated and their parametric effects on the behavior of the

process were observed The energy and mass balance model was constructed using Microsoft

excel format The rate of decarburization for the two blow scenarios (1st and 2

nd stages of

blowing periods) was also investigated to ascertain if models highlighted in literature could be

applied to the current refining process under study

33 Reduction stage after refining in the AOD

Once the desired carbon content was achieved in the refining stage (for the process under study

lt 10 wt C) the reduction stage of the decarburization process commenced to recover

chromium lost to slag as chromium oxide (Pretorius and Nunnington 2002) In this stage of the

process oxygen and nitrogen blowing was replaced by argon blowing and FeSi addition The

37

argon stirs the bath and drives out the nitrogen thereby lowering nitrogen content in the bath

(Kienel 2014) The reduction of Cr2O3 to chromium follows equation 31

211986211990321198743 + 3119878119894 = 4119862119903 + 31198781198941198742 [31]

Chromium is then recovered into the metal bath as it separates from slag It is important however

that the slag be fluid and less viscous to reduce metal entrainment in slag For this to be achieved

in the process under study fluorspar was added (as discussed in section 252 of the literature

review) to make slag more fluid (Pretorius and Nunnington 2002)

Summary

The aim for plant data collection in this section for the process under study was to measure and

investigate the effect of different process parameters such as effect of initial temperature on

decarburization process or gas flows and ratios amongst others on the overall decarburization

process The chemical analysis of the molten metal charged into the AOD was determined first

before charging the converter The weight of the molten metal was also measured and initial

temperature of bath analysed with a quick immersion thermocouple once the molten alloy was

charged into the vessel Dip thermometers and samplers were used as tools for data collection in

the process for temperature and chemical composition determination respectively All collected

data was logged onto a data template (log sheet as shown in appendix 1) up until the final metal

was tapped and cast into ingots Once all the data emanating from different process parameters

was collected analysis and modeling of data was done

38

CHAPTER 4

4 Modelling Methodology

41 Introduction

Modeling and simulation in AODs has been studied in an attempt to predict thermodynamic and

process parameters effect on the decarburization process This modeling and simulation can be

used to predict outcome of the decarburization process in the AOD making manipulation of

process parameters to achieve a desired product outcome possible (Wei and Zhu 2002) By

investigating the free energies activities of elementsoxides interaction parameters (amongst

other parameters investigated) and as discussed in literature modeling of the interaction of the

different elements and oxides it is possible to do thermodynamic modeling and parametric study

of the effect and contribution of different process parameters to the overall decarburization

process (Fruehan 1976 Wei and Zhu 2002)

A mass and energy balance model was constructed using thermodynamic principles and research

work done by authors who investigated such parameters as interaction coefficients and Gibbs

free energy amongst others This model was used as a predictive tool to determine conditions

required to achieve a desired product outcome Rate equations were also modeled based on

literature to predict final product specifications as well These rate equations were divided into

two categories namely rate equations at higher carbon contents (gt 20 wt C) and rate

equations for lower carbon contents (le 20 wt )

42 Thermodynamic analysis of plant data

Having discussed thermodynamic principles such as the activity coefficients of Cr Si C SiO2

Cr2O3 and FeO highlighted in the literature the plant data was analyzed to investigate the effect

of the thermodynamic models and rate equations on the decarburization process In addition the

slag and metal analyses were analyzed to model behaviour of the blowing process based on the

model equations formulated by Heikkinen et al (2011) that analysed behaviour of silicon

carbon and chromium in the converter

The standard free energies (ΔG) for the reactions of Cr C and Si were obtained by the values

obtained from the modelling software HSC Chemistry for Windows which provides a detailed

39

database of thermodynamic data (Xiao et al 2002) The Gibbs free energies values indicating the

change of state from Raoultian to Henrian states (ΔGRrarrH) defined as energy for the change

between Raoultian and Henrian standard states and were obtained from the formula shown in

equation 36 derived from work done by Sigworth and Elliot (1974)

ΔGRrarrH = minusnRTLnγiO [41]

The activity coefficients (γi) were obtained through calculations using the unified interaction

parameter formalism (Pelton amp Bale 2007) whereas the mass fractions (Xi) were obtained

from the chemical analyses of the slag and metal samples at different times of the blow process

in the converter The activities (ai) were then obtained by multiplying the different activity

coefficients for the slags with the appropriate mole fractions of the oxides in the slag as shown in

equation 42 (Ban-Ya amp Xiao 1993)

119886119894 = 120574119894 times 119883119894 [42]

421 Oxygen partial pressure for oxidation of C Si and Cr

The oxidation reactions of Cr Si and C were taken into account so that the oxygen partial

pressures could be determined Equilibrium conditions were assumed for the oxidation reactions

(Heikkinen et al 2011) The oxidation of silicon oxidation in the metal bath with iron as the

matrix was expressed as shown in equation 43 (Heikkinen et al 2011)

SiFe + O2(g) = (SiO2) [43]

K43 = aSiO2

aSitimesPO2

= aSiO2

γSitimesXSitimesPO2

= eminus[ΔG43

o + ΔGRrarrHRT

]

The equilibrium constant equation 43 was expressed in terms of activity of SiO2 Si and the

partial pressure of oxygen Rearranging equation 43 and expressing it in terms of oxygen

partial pressure made it possible to calculate the oxygen partial pressures for the oxidation of Si

in the converter as shown in equation 44 (Heikkinen et al 2011)

PO2=

aSiO2

γSitimesXSitimes e[

ΔG1o+ ΔGRrarrH

RT] [44]

Where aSi = γSi times XSi

40

This same calculation procedure was also followed for the other elements Cr and C and the

equations also derived as shown in equations 45-48 (Heikkinen et al 2011)

For Chromium oxidation

4

3CrFe + O2(g) =

2

3(Cr2O3) [45]

K45 = aCr2O3

23

aCr4 3frasl

timesPO2

= aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl

timesPO2

= eminus[ΔG45

o + ΔGRrarrHRT

]

PO2=

aCr2O3

23

γCr4 3frasl

timesXCr4 3frasl times e[

ΔG45o + ΔGRrarrH

RT] [46]

For carbon oxidation

2CFe + O2(g) = 2CO(g) [47]

K47 = PCO

2

aC2 timesPO2

= PCO

2

γC2 times XC

2 times PO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

PO2=

PCO2

γC2 timesXC

2 times e[ΔG1

o+ ΔGRrarrHRT

] [48]

The data collected was then taken and the oxygen partial pressures calculated for each heat

422 Equilibrium Partial pressure of CO from oxidation of C Cr and Si

The equilibrium partial pressures for carbon monoxide for the set of plant data were also

determined In this case the reduction reactions for SiO2 and Cr2O3 to elemental Si and Cr

respectively were taken (Heikkinen et al 2011) The reduction reactions were taken at

equilibrium and the equilibrium constants were then rearranged in terms of the carbon

monoxide partial pressures In addition the oxidation of C to form CO was also taken into

account and the thermodynamic relationships are shown in equations 49- 411

For silica reduction

2C + SiO2 = Si + 2CO [49]

K = aSitimesPCO

2

aC2 timesaSiO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Rearranging equation 49 gives

41

PCO = radicaC

2 timesaSiO2

aSitimes eminus[

ΔG1o+ ΔGRrarrH

RT] [410]

For chromium oxide reduction

3C + Cr2O3 = 2Cr + 3CO(g) [411]

K = γCr

2 timesPCO3

aC3 timesaCr2O3

= eminus[ΔG1

o+ ΔGRrarrHRT

]

Combining equations 49 ndash 411 gives

PCO = radicγC

3 timesaCr2O3

aCr2 times eminus[

ΔG1o+ ΔGRrarrH

RT]

3

[412]

Carbon oxidation by gaseous oxygen

2C + O2 = 2CO [413]

K = PCO

2

aC2 timesPO2

= eminus[ΔG1

o+ ΔGRrarrHRT

]

The partial pressure of CO gas is then given by

PCO = radicaC2 times PO2

times eminus[ΔG1

o+ ΔGRrarrHRT

] [414]

From the values that will be obtained for the partial pressure of oxygen will determine the order

of oxidation of Cr Si and C in the metal bath For the oxidation reactions (of Si C and Cr) to be

in mutual equilibrium with each other the calculated values of the partial pressures of oxygen

will have to be the same or close in comparison (Heikkinen et al 2011) The calculated oxygen

partial pressures were compared for higher and lower carbon contents as discussed by Heikkinen

et al (2011) who stated that the partial pressure for oxygen increases as the carbon content in the

metal bath decreases

43 Rate equations in the converter decarburization process

As discussed in sections 2581 and 2582 of the literature review oxidation of carbon by

Cr2O3 at higher carbon contents is determined by the rate of oxygen being blown into the metal

bath and by the liquid mass transfer of carbon to the bubble surface at lower carbon levels

(Fruehan 1976) As a result of this there are two distinct regimes of carbon oxidation in the

AOD for higher and lower carbon contents in the metal bath (Wei and Zhu 2002)

42

431 Rate equations at high carbon contents

According to Wei and Zhu (2002) at high carbon contents in the metal bath the average losses

of the different components (Silicon Carbon and Chromium in this case) can be modelled

separately according to the rate equations below For the purpose of this study the terms high

carbon and low carbon levels were taken as defined as C ge 2 wt and lt 2 wt respectively

In this case the equations derived by Wei and Zhu (2002) were utilized to determine the rate of

oxidation of C Cr and Si at high carbon contents

d[C]

dt= minus

2ηQO

22400times xC times

100lowastMC

Wm [415]

d[Cr]

dt= minus

2ηQO

22400times xCr times

100lowastMCr

15lowastWm [416]

d[Si]

dt= minus

2ηQO

22400times xSi times

100lowastMSi

2lowastWm [417]

Where

η is the utilization ratio

QO is the flow rate of oxygen cm3s

-1

Mi is molar mass of substance (i) expressed in gmol-1

Wm is the mass of the liquid metal bath in grams

Xi is the distribution ratio of oxygen within each component (i) in the liquid metal bath

The distribution ratios of blown oxygen among the elements (Xi) are proportional to the Gibbs

free energies of their oxidation reactions at the interface (Wei and Zhu 2002) The equations

proposed by Wei and Zhu (2002) were also applied to the data collected from the plant

xC = ΔGC

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [418]

xCr = ΔGCr

3frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [419]

xSi = ΔGSi

2frasl

ΔGC+ ΔGCr

3frasl + ΔGSi

2frasl [420]

Where

ΔGi is the Gibbs free energy for oxidation reactions of element (i) in Jg-1

xi is the distribution ratio of blown oxygen for element (i)

43

432 Rate equations for low carbon levels

The rate equation at low carbon content was derived from the work done by Wei and Zhu (2002)

as shown in equations 421 ndash 424 For the purpose of this study low carbon contents were

defined as carbon le 2 wt and this in accordance with the definition of low carbon content

used in the process under study

Wei and Zhu (2002) stated that in the refining process oxidation of elements in the metal bath or

reduction of oxides occur at the bubble surface and hence the area of the interfacial reaction

(Area) is the approximate total area of the bubbles in the bath The estimation of the area of

interfacial reaction (Area) was defined as the approximate total surface area of the bubbles that

is available for oxidation or reduction of elements during refining (Wei amp Zhu 2002) Using the

expression proposed by Diaz et al (1997) the estimation for the total number of bubbles

becomes

nb = 6QHb(πdb3μb) [421]

Where

nb The total number of bubbles

Q Total gas flow rate cm3s

-1

Hb Rising height of bubble

db Average diameter of bubble

μb Velocity of bubble cms-1

The velocity of the bubbles in this case is then expressed as shown in equation 422 (Davies and

Taylor 1950 Wei and Zhu 2002)

μb = 102(gdb

2)

12frasl [422]

Based on the relationship shown in equation 421 and 422 the estimation of the area of reaction

interface can be derived as (Wei and Zhu 2002)

Area = 6QHb(dbμb) [423]

44

The numerical value of Hb used in the present study was adopted from work by Wei and Zhu

(2002) where it was calculated to be 95cm At low carbon content the liquid mass transfer of

carbon in the bath becomes rate limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999

Fruehan 1976) The mass transfer coefficient of carbon in the liquid bath (kc) (Baird and

Davidson 1962) was calculated as shown

kC = 08reqminus1 4frasl

DC1 2frasl

g1 4frasl [424]

Turkdogan (1980) stated that at a sufficiently high gas stream velocity the bubble sizes are

generally uniform From the work done from the water modelling work done by Wei at el

(1999) Dc and db were adopted from the figures utilized by Wei and Zhu (2002) and being 746

cm2s and 25 cm respectively as applied

433 Equilibrium concentration of carbon

The typical reactions taking place during the refining process involves the oxidation of Cr and C

and being exothermic reactions the temperature of the bath is consequently raised (Wei and

Zhu) The equation below was then appropriately approved

(Cr2O3) + 3[C] = 2[Cr] + 3CO [425]

The equilibrium concentration of carbon at the interface was obtained by the formula shown

below

[C]e = PCO

γCradic

a[Cr]2

aCr2O3times KCrminusC

3

[426]

The partial pressure used in the above equation was obtained from each of the 30 plant data

samples collected at the second stage of the blowing process The results were then tabulated so

that they could be compared to the values obtained from plant measurements to determine if the

model followed same trend as plant data

45

44 Mass and energy balance for the AOD process

The input data for the model was tabulated as shown in appendices The blowing time for each

of the blowing intervals was also specified for each stage and was manipulated to find optimum

time per specific flow rate to achieve the desired carbon level

The model took into account all the materials charged into the AOD that is burnt lime burnt

dolomite FeSi and the molten alloy from the EAF Each of the materials was analysed

compositional consistency in order to calculate the input moles of the respective elements and

oxides into the process

The initial temperature of the bath was also measured as this is critical in thermodynamic

calculations of changes in the Gibbs free energy of reactions (ΔG) The thermodynamic

calculations were conducted based on the thermodynamic data obtained from HSC version 70

(H-enthalpy S-entropy C-heat capacity) thermodynamic software

The molar fractions and energy contributions to the process were calculated for the input

materials The elemental constituents of the alloy were analysed using the Spectromaxx Optical

Emission Spectroscopy Analyserreg while the burnt lime and dolomite were analysed using the X

Ray Fluorescence (XRF) analysis method As the burnt lime dolomite and FeSi were being fed

into the AOD from the feed bunkers initial temperature of 25oC was assumed for the material

The initial analysis and temperature in this model was selected randomly to form the basis for

calculations

The initial alloy content was added as one of the inputs To determine decarburization extent the

initial carbon content had to be obtained For the purposes of the model derivation the analysis

used for calculations was randomly chosen The weight was also recorded to assist in

determination of the contributions of each element and the temperature to calculate Gibbs free

energies for each element in the metal bath

Burnt lime and dolomite analysis was also taken into account Though in this case the CaO

content for burnt lime was taken to be 100 wt the model allows for adjustments according to

analysis and specification of raw materials to be used The initial temperature for these raw

46

materials was taken as 25oC as the material charged into the AOD came from storage bunkers at

room temperature

The ratio of addition of lime dolomite of 15 was used Dolomite is added in the

decarburization process to enhance the kinetics when the process goes to the reduction phase

The other advantage of dolomite addition is that the MgO in the slag reacts with silica (SiO2)

forming an intermediate phase of CaMg silicate and these phases are low melting thus making

the slag quite fluid and eliminating need for fluorspar enhancing mass transfer processes due to

reduced viscosity and consequently improve chromium recovery from slag (Nunnington 2002)

The model also catered for ferrosilicon addition for chromium recovery from slag

On the inputs spreadsheet the input data was used to calculate the mass (in kg and kmols) of the

individual elements and oxides The detailed calculations are shown in the methodology chapter

Thermodynamic considerations were used to calculate predicted output and contribution of each

element or oxide to the overall mass and energy balance

The process outputs were also calculated using the same formulae as for process inputs (as

described in the methodology) In this case a carbon composition in the final product of 10 wt

was targeted and fixed The model however utilizes excel solver to iterate to a predicted metal

and slag composition based on the calculations already described

The slag quantity generated from the decarburization was also calculated using the quadratic

formalism The CO produced was a function of the C oxidized from the liquid bath as shown in

the calculations Once completion of model was achieved comparison with results obtained

from Wei and Zhursquos rate equations could be compared to the plant data collected The values of

activity coefficients and activities obtained from the mass and energy balance model were also

obtained and tabulated (appendix 10 11 12 and 13)

441 Mass and energy balance construction procedure

The energy and mass balance model was constructed for the current AOD process The table 41

shows the input data that was used The (model was based on the first law of thermodynamics)

principle of energy in = energy out (also implying to mass) is the basic law of conservation of

energy that was applied to the model Use of thermodynamic principles and work done by

researchers such as Heikkinen et al (2011) Fruehan (1976) Ban-Ya (1992) amongst others was

applied into derivation of thermodynamic behaviour of the process under study Assumptions to

47

certain process parameters such as initial temperature and weight of molten alloy were done to

allow for calculations to be carried out The model however allowed for change of these

parameters to calculate any different scenario presented Table 41 shows analysis of the initial

raw materials used

Table 41 Analyses of Material inputs into the AOD

Lime Dolomite FeSi Metal Lime

Dolomite - 15

Al2O3

Mass 0 Al2O3 Mass 000 Al Mass 000

Mass of

Tapt 7

CaO

Mass 80 CaO Mass 2000 C Mass 000 Analysis C 3

Cr2O3

Mass 0 Cr2O3 Mass 000 Cr Mass 000 Cr 67

MgO

Mass 2 MgO Mass 3000 Fe Mass 6627 Si 030

FeO

Mass 0 FeO Mass 000 Si Mass 3373 Fe 2970

SiO2

Mass 0 SiO2 Mass 000

LOI

(CO2)

Mass 0 LOI

(CO2) Mass 5000

For the initial calculation a single tap was taken and the inputs analysed as shown above From

this data the kmols and overall energy were calculated as shown below

441 Calculating the initial mass (moles) and energy of the metal bath in the converter

The initial mass (in moles) and the energy of the metal bath were calculated based on equation

427

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119898119890119905119886119897 [427]

In order to get the moles of each element (in kilo-moles) the relationship between the mass of

each element and its molar mass was considered as shown

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [428]

All calculations were taken at 1600 oC and ΔH also taken at the same temperature The energy

for each element was then calculated as shown below

48

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 1600119900119862) [429]

The same calculation was done for the main elements in the molten metal and tabulated as

shown in table 42

Table 42 Moles and Energy for metal 1600 oC

Mass Masskg kmol mole MJ

C 30 210 1748 12 56572

Cr 670 4690 9020 62 491185

Si 03 21 075 1 6827

Fe 297 2079 3723 26 280567

Total 100 7000 14566 100 835151

442 Calculating the mass (kmols) and energy balance for lime and dolomite

The burnt lime and dolomite added into the converter just after charging of the molten alloy is

for slag formation in the vessel The targeted slag basicity is approximately 18 ((Cao +

MgO)SiO2) and the chemistry is 40 wt CaO 15 wt MgO and 30 wt SiO2 as discussed

in literature

On calculating the kilomols and Energy for burnt lime and dolomite as added into the converter

for slag formation the calculations shown below were done

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904 (119897119894119898119890) [430]

But total lime is obtained as shown in equation 431 as shown

119879119900119905119886119897 119897119894119898119890 = 119905119900119905119886119897 119889119900119897119900119898119894119905119890 times 119872119886119904119904119862119886119874 (119894119899 119897119894119898119890) [431]

The total dolomite (requirement) was calculated as shown in equation

119879119900119905119886119897 119889119900119897119900119898119894119905119890 =119905119900119905119886119897 119872119892119874 119898119886119904119904 (119894119899 119904119897119886119892)

119872119886119904119904 119872119892119874 (119894119899 119889119900119897119900119898119894119905119890) [432]

49

In order to calculate the total mass of MgO in slag the activity of MgO in the slag had to be

determined and then the total MgO in slag then determined by the expression shown in equation

433

119905119900119905119886119897 119872119892119874 (119894119899 119904119897119886119892) = 119886119872119892119874 times 119905119900119905119886119897 119904119897119886119892 119898119886119904119904 [433]

The total slag mass was obtained as shown in equation 334 (XRAM Technologies 2016)

MassSlag = Mols(FeSi+Alloy)Cr minus

Mass(Cr)

Mass(Si)lowast

Mr(Si)

Mr(Cr)timesMols(Alloy+FeSi)Si

(2times

aCr2O3MrCr2O3

minus MrSitimesaSiO2timesaCr

MrSiO2timesaSitimesMrCr

)

[434]

aCr2O3 ndash Activity coefficient for Cr2O3

aSiO2 ndash Activity coefficient for SiO2

Mri ndash Molecular Weights for element i

aSi ndash Activity for Si

aCr ndash Activity for Cr

The energy and mols for CaO also calculated with the same formulae as was done for the metal

elements The calculation for the activity coefficients was also done and will be shown later

Energy calculations were done using ΔH at 25oC

The same calculations as done for lime were implemented were implemented for dolomite As

shown on the burnt lime calculations the mass of slag is calculated same way Energy

calculations were also done using ΔH at 25oC

119872119886119904119904119862119886119874 = 119872119886119904119904119862119886119874 times 119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 [435]

From equation 435 total Dolomite mass was calculated as

119879119900119905119886119897 119898119886119904119904119889119900119897119900119898119894119905119890 = 119872119886119904119904 119900119891 119872119892119874 119894119899 119904119897119886119892

119872119886119904119904119872119892119874 119894119899 119889119900119897119900119898119894119905119890 [436]

50

From the equation 436 above mass of MgO in dolomite was calculated as shown in equation

437

119872119886119904119904119872119892119874 = 119886119872119892119874 times 119872119886119904119904119878119897119886119892 [437]

443 Kmols and energy calculations for FeSi in the reduction stage

Calculations for moles mass and energy were done for Cr Fe and Si using the same calculations

and equations as for the alloy The difference is in the energy calculations as value of ΔH used

was at 25oC

119872119900119897119886119903 119898119886119904119904 = 119872119886119904119904 (119894119899119894119905119894119886119897 119897119894119902119906119894119889 119898119890119905119886119897) times 119872119886119904119904 119891119903119886119888119905119894119900119899 119900119891 119890119897119890119898119890119899119905 119894119899 119865119890119878119894 [438]

To obtain the moles of each element the relationship between the mass of each element and its

molar mass was considered as shown in equation 439

119870119898119900119897119904119883 = 119872119886119904119904 119900119891 119890119897119890119898119890119899119905 119894119872119903119894 [439]

All calculations were taken at 25 oC (ambient temperature at which FeSi was charged into the

converter) and ΔH also taken at the same temperature The energy for each element was then

calculated as illustrated in equation 440

119864 = 119896119898119900119897119904119894 times ∆119867119894(119886119905 25119900119862) [440]

Calculations done for the FeSi were tabulated in table 43

Table 43 Moles and Energy calculations for FeSi

FeSi

Mass Masskg kmol mole MJ

Fe 34 8434 300 5031 000

Si 66 16566 297 4969 000

100 25000 597 10000 000

51

444 Mass and energy balance calculations for AOD gases

For the process under study argon was only used in the reduction phase of the process In the

model however provision was made for Argon in case there would be change in process and

Argon used in place of nitrogen The gas flow ratios of oxygen and an inert gas for this model

give allowance for either nitrogen or argon to be used in conjunction with blown oxygen The

gases used in the model are all pure gases as are the gases used in the process under study

4441 Oxygen

The mass of oxygen was taken as 100 because it was pure Oxygen Therefore the total

mass of oxygen can then be calculated as shown in equation 441

Total Mass O2 =XminusY

2timesMrO2

[441]

Where

X is mols of oxygen in the oxides in slag and oxygen in the gas produced as a result of the

oxidation reactions (namely CO) and is calculated as

X = 3 times (MolAl2O3+ MolCr2O3

)slag + 2 times ((Mols(slag) + Mols(gas))SiO2) + ((MolsCaO + MolsFeO + MolsMgO)slag +

MolsCO(gas)) [442]

And Y is the molar contribution of the oxygen from the slag formers (lime and dolomite) and is

calculated as

Y = 3 times ((MolsLime + MolsDolomite)Al2O3+ (Molslime + Molsdolomite)Cr2O3

) + 2 times (MolsSiO2+ (MolsSiO2

+ MolsCO)Dolomite) + ((MolCaO + MolFeO + MolMgO)lime + (MolCaO + MolFeO + MolMgO)Dolomite

[443]

In summary the oxygen mass supplied through blowing into the metal bath could be obtained by

subtracting the total oxygen in the slag (in the form of oxides) and the oxygen supplied within

the oxides in the inputs (lime and dolomite etc) This model however assumes that all oxygen

utilized in oxidation reactions and no free oxygen escapes from the bath unreacted The model

does not also take into account post combustion reactions All reactions are assumed to occur

and finish in the decarburization interval

52

4442 Argon

In the case where argon is blown in conjunction with oxygen typical oxygenargon gas ratios are

basically 31 at higher carbon contents and 11 at lower carbon contents for oxygen and argon

respectively during the decarburization process The argon used in the model is pure argon

Mass = 100 wt

Mol = 100 wt

However equation 441 was used in case when crude argon would be used in the process

119872119886119904119904 119900119891 119860119903119892119900119899 = 119872119886119904119904 times 119879119900119905119886119897 119872119886119904119904 119900119891 119860119903119892119900119899 [444]

But

Total mass (Ar) = Flow rate(Ar)

flow rate (O2)times total Mol (O2) times Mr(Ar) [445]

The total flow rate of Argon was then obtained as shown in equation 443

Total Flow rate (Ar) = flow rate(O2)

O2Arfrasl Ratio

= sumflow rate (O2)

O2Arfrasl ratio

yn=1 [446]

Where n = 1-y blowing stages

For these particular calculations Argon was not blown during the decarburization stages

Table 44 Moles and Energy calculations for Argon at 25 oC

Argon

Mass Masskg Kmol mole MJ

100 0 0 100 0

4443 Nitrogen

Nitrogen is the inert gas used in the process under study in the AOD converter The other reason

for the choice of nitrogen over argon is the fact that nitrogen is significantly cheaper than argon

to purchase The total nitrogen requirement is calculated as

Mass N2 = Mass times total N2mass [447]

53

But the total N2 mass is calculated by taking into account the gas flow ratios of nitrogen and

oxygen and the molar masses of nitrogen and oxygen as shown in equation 448

Total N2mass =N2flow rate

O2flow ratetimes total Mols(O2) times Mr(N2) [448]

But the total flow rate is expressed as shown in equation 449

Total Flow rate N2 =total flow for all stages (O2)

total O2

N2frasl

[449]

Total Flow rate N2 = sumflow (O2)n

O2N2

frasl ration

yn=1 n=1-y

Y is defined as the number of blowing stages in the decarburization process with differing gas

flow ratios

45 Thermodynamics and Equilibrium considerations in AOD refining process

Equations 29 ndash 215 as described in the literature show competing reactions occurring in the

converter The general equation for the oxidation of carbon is shown in equation 450

Taking the following equations 29 ndash 211 from literature and applying a general metal reduction

equation equation 450 becomes

pMxOy + zC = rCO + sM [450]

At equilibrium equation 450 can be expressed in terms of the equilibrium constant (K) as

illustrated in equation 451

K = aM

s timesPCOr

aMxOy

ptimesaC

z [451]

Taking into account the Gibbs free energy equation 451 can be further expressed as shown in

equation 452 below

K = aM

s timesPCOr

aMxOy

ptimesaC

z = eminus∆G+ ∆GRrarrH

RT [452]

Where ΔG is the free energy of reaction

54

ΔGRrarrH free energy for change between Raoultian and Henrian

T is the temperature measured as K

R is the universal gas constant and has value of 8314 Jmol

Pi partial pressure of the component

ɑi is the activity of the components

But αi = γiXi

Pi = XiP

With the preferred use of nitrogen as the inert gas for the process under study it is paramount to

investigate nitrogen solubility in the metal bath during the decarburization process Fruehan

(1992) stated that nitrogen dissolution has significantly higher in ferrochromium alloys than any

other steel He further stated that an increase in chromium content increases nitrogen dissolution

whilst an increase in sulphur decreased nitrogen dissolution From the investigation Fruehan

(1992) carried out he concluded that for the dissolution of nitrogen followed equation 453

below

1

2N2g

= N [453]

K = PN

PN2

12

= eminus∆G+∆GRrarrH

RT

46 Interaction coefficients in metal and slag in the refining process

Wada and Saito (1951) described interaction parameters relating to energy of mixing hence

dependent on interaction energies between solute and solvent atoms or between the solute atoms

themselves The authors further stated that the interaction parameters also related to the energy

of mixing Interaction coefficients were calculated for both metal and slag phases In

determining and estimating activity coefficients in this model certain assumptions were made

based on the limits highlighted in the points below

The average temperature of the bath during the initialstarting stages of the refining

process was taken to be 1600oC As a result the interaction coefficients were calculated

at 1600oC The assumption was that the variation of the interaction coefficient values was

not significant with the temperature variation in this study

55

The interaction coefficients proposed by Sigworth and Elliot (1974) are for dilute

solutions in liquid iron as solvent and their applications to Fe-Cr-C melts is based on

low Cr contents

With the Nitrogen solubility model Kim et al (2009) developed it for FeCr with Cr 15-

60 and partial pressure of N2 between 004-097 atm It should be noted that the Cr

content in the Fe-Cr-C melts considered in the process under study was higher than the

15-60 wt highlighted and could be as high as 73 wt in the metal

451 Activity coefficient calculations in metal phase

As discussed in literature the refining process of a Fe-Cr-C system and the co-existence of

solute atoms in a liquid bath were investigated An increase in chromium content will result in a

decrease in carbon activity (Ohtani 1956) The activity coefficients for the metal phase

constituents were calculated according to studies by Pelton and Bale (1986) who derived the

following expression to determine the individual activity coefficients

lnγsolvent = minus1

2sum sum εjk

Nk=1 XjXk

Nj=1 [454]

lnγi = lnγSolvent + sum εijXjNj=1 [455]

From the expressions shown in equations 454 and 455

X ndash Molar fraction of the component i or j

εij Is the interaction parameter for components i and j and was obtained by the expression

shown in equation 456 (Sigworth and Elliot 1974)

εij = 230 timesMWi

MW1times σi

j+

MW1minusMWj

MW1 [456]

MWi is the molar weight of a component i

MW1 is the molar weight of the solvent

σ is the 1st order interaction parameter

From equations 454 ndash 456 the activity coefficients for the different elements in the alloy bath

were calculated and tabulated as shown in table 45

56

Table 45 Activity Coefficients for each element i in the ferrochromium alloy

Mass Mole

Activity

Coeff Activity

Al 000 000 144 000

C 100 393 008 000

Cr 6554 5943 091 054

Fe 2993 2527 112 028

N(g) 323 1087 002 000

Si 030 050 175 001

Table 46 First Order Interaction Coefficients for liquid iron (Sigworth and Elliot 1974)

First order interaction coefficients in liquid iron

Item Al C Cr N Si

Al 00450 00910 - - 0058 00056

C 00043 01400 - 0024 01100 00800

Cr - - 01200 - 00003 - 01900 - 00043

N - 0028 01300 - 0047 - 00470

Si 00580 01800 -00003 00900 01100

According to Kim et al (2009) the nitrogen solubility in the metal bath can be expressed as

shown in equation 457 and that was the model used to predict nitrogen solubility in the

investigation

logγ = (minus148

T+ 0033) times XCr + (

156

Tminus 000053) times XCr

2 [457]

XCr is the mass of Cr

T is the operating temp i

The mass of Cr was taken from the final product

452 Activity coefficient calculations in the Slag Phase

From the work done by Ban-Ya (1992) it was concluded that by taking a component (i) in a

multi component slag the activity coefficient of this component can be expressed in terms of the

quadratic formalism The author went on to state that even for liquid silicate melts that are not

strictly regular solutions the same quadratic formalism could be used as if these solutions were

57

regular solutions For slag phase components the model by Ban-Ya (1993) was used The model

is illustrated in equation 458 as shown below

RTlnγi = sum aijXj2N

j=i+1 + sum sum (aij + aik minus ajk)XjXkNk=j+1

Nj=i+1 [458]

Where ɑ is activity of an oxide jik

R ndash Universal gas constant 8314Jmol

T Temp measured in degrees Kelvin

ΔG ndash Gibbs free energy of reaction ndashJmol

ΔGR-gtH ndash Gibbs free energy for change between Raoultian and Henrian standard states

The calculations for mole fractions for the oxides in the slag were calculated in order to

determine the activity coefficients above Equation 459 illustrates the molar fraction calculation

for CaO

MassCaO = MolesCaOMrCaO

sum (MrtimesMoles)ini=1

[459]

But the molar percentage of CaO in the slag is expressed as

MolesCaO = MolesSiO2times Molar Basicity

CaO

SiO2 [460]

Where the molar basicity of slag is expressed as a ratio of CaO and SiO2 as shown in equation

461 below

Molar BasicityCaO

SiO2= Required Basicity times

A+BlowastC

D+A+Btimes(C+E) [461]

A =(Mass

Mrfrasl )CaO

sum (Mass)ini=1

for dolomite

B =(

Lime

Doloratiotimessum (

Mass

Mr)

i

ni=1 +

Lime

Doloratiotimes(1minus

(Total Mass)LimeMrCO2

))

sum (Mass

Mr)t

nt=1 +(1minus

sum (mass)ana=1

MrCO2)

C =lime

doloratiotimesMassCaO

Lime

Doloratiotimessum (

Mass

Mr)p

np=1

in lime

D =(

Mass

Mr)MgO

sum (Mass

Mr)z

nz=1

in dolomite

58

E =Lime

Doloratiotimes(

Mass

Mr)MgO

lime

Doloratio sum (

Mass

Mr)b

nb=1

in lime

The same equations can be used for MolesMgO except for C which then becomes as shown in

equation 462

C =lime

doloratio

MassMgO

MrMgOin dolomite [462]

The table below shows values obtained from the calculations

Table 47 Interaction coefficients for slag phase

Mass Mole Activity Coeff Activity

Al2O3 000 000 001 000

CaO 051 048 034

Cr2O3 001 000 878 004

FeO 014 011 143 016

MgO 008 012 013

SiO2 030 028 016 003

Summary

In this chapter thermodynamic concepts such as interaction coefficients were determined for the

creation of the mass and energy balance model for the decarburization process Interaction

parameters are a measure of the interaction of an element or a compound with neighbouring

atoms or compounds Determination of activities and activity coefficients for individual

components in the alloy bath and for each of the components in the multi-component slag melt

was done in order to construct the mass and energy balance modelrsquos mass balance Rate

equations were also modelled mainly based on work by Wei and Zhu (2002) amongst other

researchers Once the models were constructed they were then used as analysis tools for plant

data and predictive purposes From the rate equations it became possible to predict final carbon

content for the two different blowing stages under study namely high and lower carbon blowing

conditions Tabulated figures for parameters used in the calculations and results obtained in the

mass and energy balance model for input and output summaries are found in appendices 2 ndash 16

59

CHAPTER 5

5 RESULTS AND DISCUSSION

51 Introduction

The previous chapter 3 and 4 on methodology were based on process description and model

constructions based on thermodynamic derivations and rate equations In this chapter the

process plant data was applied to the research done to investigate the effect of individual process

parameters on the process under study Different relationships and effect on overall process

performances were investigated as well to understand how these relationships also impacted on

the decarburization process

52 Carbon removal trend with decarburization for plant data

Plant data for 9 heats were randomly taken and the carbon measured based on the initial carbon

content against the carbon content during the course of the oxygen blowing stage and at the end

of the blow The intervals of sampling were largely determined by the AOD operator as well as

the conditions experienced during the decarburization process

In this context the initial carbon content refers to the carbon in the alloy bath after tapping from

the EAF and charged into the AOD for the decarburization process The results showing the

change in the carbon content as a function of time were then plotted in order to ascertain the

changes in carbon content with time Generally the change in the carbon content of the bath

followed a polynomial curve as shown in in Figure 51

60

Figure 51 Change in carbon content of the bath as a function of blowing time for 9 random heats

As shown in Figure 51 the rate of decarburization was quite significant as characterized by a

steep decrease in carbon content from the initial 781 wt to 304 wt from the onset of

blowing up to around 50 mins into the blow for the average trend This is characterized by the

steep decrease in carbon content observed on the graph above According to Wei and Zhu

(2002) the mass transfer of carbon to the reaction interface is not rate limiting at high carbon

contents and as a result the overall rate of decarburization tends to be limited by the rate of

oxygen blowing into the bath andor the rate of mass transfer of oxygen to the reaction interface

In this regard the observed carbon removal efficiency (CRE) also tends to high during the initial

stages of the blow (Wei and Zhu 2002 Turkdogan and Fruehan 1999)

The observed in the carbon content tends to level off after 50 mins into the blow In other words

rate of change in the carbon content drops significantly as the carbon levels approaches critical

carbon content after which the mass transfer of carbon to the reaction sites becomes rate

limiting (Wei and Zhu 2002 Turkdogan and Fruehan 1999) At this stage the rate of

decarburization becomes significantly slower and the observed trend on the graphs continue to

level off towards the end of the blowing process After the attainment of critical carbon content

the oxidation of chromium start to become significant and as a result chromium losses to slag

are experienced (Wei and Zhu 2002) Based on the AOD converting process of stainless steel

Visuri et al (2013) estimated that the chromium oxide (Cr2O3) content of slags usually increases

to about 30 to 40 wt at the end of the decarburization stage

0

1

2

3

4

5

6

7

8

9

0 50 100 150 200

C

arb

on

Timemins

Ave

61

From the plant data graphs showing the change in carbon content (ΔC) ie the difference

between the initial and final carbon contents as a function of blowing time for both 1st and 2nd

stage blows were constructed to evaluate if there was any relationship between the blowing time

and the amount of carbon removal from the bath The rate of change of carbon content (ΔC)

was further plotted as a function of time and the results are shown in Figure 52

Figure 52 Drop in carbon content with blowing interval

It was observed for the longer time intervals resulted in a bigger drop in carbon composition

during the first stage of the blow This meant that the longer blowing cycles the lower the

residual carbon content of the bath due to more time available for mass transfer of carbon thus

more carbon exposed for oxidation (Wei and Zhu 2002 Fruehan 1976)

The second blow stage showed an inverse trend The longer the blow was the lower the change

in carbon in the bath This is illustrated in the graph shown in Figure 53 The graph shows a plot

of the second blow duration for each heat and the change in carbon (ΔC) observed from the

plant data

4

45

5

55

6

65

7

75

0 50 100 150 200

C

Blowing intervallength for 1st stage- mins ΔC Linear (ΔC)

62

Figure 53 Change in carbon content (ΔC) with blowing time

At lower carbon contents decarburization becomes more difficult particularly as the carbon

content in the bath approaches the critical carbon range and the loss of chromium due to

oxidation becomes significant (Wei and Zhu 2002 Turkdogan and Fruehan 1999 Visuri et al

2013) Furthermore the chromium affinity for carbon also makes it difficult to decarburize at

low carbon contents As discussed in literature chromium has affinity for carbon and mainly

exists as chromium carbides in the high carbon ferrochromium alloy (Gasik 2013 Ohtani 1956

Venugopalan 2005)

As a result the affinity of chromium for carbon was investigated to evaluate if it had any impact

on the decarburization process The ratio of chromium to carbon (CrC) was plotted with the

rate of decarburization for the first stage of the blow at higher carbon content and for the

second stage blow at lower carbon content in the metal bath It was observed that for both the

first and second stages of the decarburization process the ratio of CrC generally increased

as the rate of decarburization decreased implying that it became more difficult to remove the

carbon as the chromium content increased and as the carbon content decreased (as shown in

figures 54 and 55)

The CrC ratio is lower for the first decarburization stage than for the second stage due to the

higher carbon composition in the first stage of decarburization From the figures 54 and 55 the

higher the CrC the lower the ΔC hence the lower the carbon content removed from the

alloy melts As the carbon content in the alloy decreases the CrC ratio increases as

0

05

1

15

2

25

0 20 40 60 80 100 120 140

Δ

C

Time interval for 2nd blow duration mins ΔC Linear (ΔC)

63

illustrated by the difference in CrC values between figures 54 and 55 The second stage

blow (as represented by figure 55) is usually characterized with the region of critical carbon

where carbon removal becomes more difficult and hence an increase in the chromium losses

This can also be attributed to the chromium affinity to carbon and as the carbon depletes in the

bath the chromium affinity for oxygen becomes significant (Gasik 2013 Wei and Zhu 2002)

The figure 54 below shows the relationship between CrC ratio and ΔC and the trend shows

that a decrease in the CrC ratio results in a higher carbon removal

Figure 54 Relationship between the CrC ratio and rate of decarburization during the first stage of blow

Figure 55 at lower carbon contents as is characteristic of the second blowing stage also shows

the same trend as figure 54

Figure 55 Relationship between CrC ratio and rate of decarburization during the second stage of the blow

80

85

90

95

100

105

000 100 200 300 400 500 600

C

r

C

ΔCdt CrC

000

1000

2000

3000

4000

5000

6000

7000

000 020 040 060 080 100 120 140 160

C

r

C

ΔCdt CrC Linear (CrC)

64

53 Mass balance for AOD refining stage

In summary the mass and energy balance was constructed as an analysis and predictive tool to

analyse medium carbon ferrochromium production in the AOD Thermodynamic principles were

used as basis to predict behaviour of elements and oxides in the AOD bath This included

determination of Gibbs free energy of the elements at different temperatures and calculation of

activities and activity coefficients The rate equations as discussed in chapter 4 were also used

to determine decarburization rates in the first and second stages of blowing corresponding to

higher and lower carbon contents respectively The performance of both models (mass and

energy balance and rate equation models) was then compared against the achieved process plant

data to measure the effectiveness of each model as a predictive tool

Table 51 shows a summary of the principles used in the calculation of the energy and mass

balance model for the AOD process under study Quadratic formalism by Ban-Ya et al (1993)

was used in conjunction with slag analysis obtained from samples taken during the blows and

the data was used to calculate the activities of the oxides in the slag The models proposed by

Pelton and Bale (1966) on unified quadratic formalism and by Sigworth and Elliot (1974) on the

thermodynamics of liquid dilute iron alloys were used to calculate the activity coefficients the

individual elements in the alloy

Table 51 Summary of models used in the calculations of mass balance in the AOD

Temp

Temperatures were measured with a dip thermocouple during decarburization

Activities of SiO2 amp Cr2O3 1198861198781198941198742

11988611986211990321198743

Quadratic formalism (Ban-Ya et al) with measured slag analysis from the process

Activity coefficients of Cr Si amp C γSi γC γCr

(Pelton amp Bale (1966) amp Henrian standard states for different metal analyses obtained from samples taken throughout the process

Xi XSi XC XCr Masses and metal sample analyses taken during the blowing down of carbon

∆119866119894119900 ∆119866119878119894

119900 ∆119866119862119900 ∆119866119862119903

119900 Calculated from HSC simulation software

Signifies change from standard to Raoultian conditions Obtained from the equation RTlnγi

PCO

Signifies the calculated estimate of ferro-static pressure in the AOD

65

531 Decarburization process at high carbon content during the first stage of the blow

On completion of the two models it became possible to compare their performance according to

the plant data obtained The modelsrsquo applicability was ascertained by keeping conditions the

same and comparing output to the plant data

5311 Carbon removal comparison between models and plant data

The results from the models done one using the Wei and Zhu (2002) rate equations at higher

carbon and the other model formulated from heat and energy balance were tabulated and shown

in Figure 56

Figure 56 Comparison of model results vs plant data for the first stage of the blow

The model constructed was used to predict the conditions of flow that would lead to the

attainment of the same carbon level obtained in the plant under the same time intervals By

taking the time interval taken for each stage of the blowing process initial temperature and metal

analyses the same as in the process plant data collected the gas flow rates (for oxygen and

nitrogen) were varied until their values produced same carbon level output as observed in the

plant data It was also observed that the shorter the time interval the higher the gas flow rates

that are required to oxidize the carbon content down to the desired final composition

A ratio of 21 for oxygen and nitrogen respectively was used for the first blowing stage The

basis of this assumption was based on work by Fruehan (1976) and Wei and Zhu (2002)

000

100

200

300

400

500

600

700

800

900

0

05

1

15

2

25

3

35

0 5 10 15 20 25 30

Rat

e E

qu

atio

n

C

Mo

de

l amp p

lan

t d

ata

C

no of plant heats Model Plant Rate Eqn

66

According to these researchers the mass transfer of mass transfer to reactions sites is not rate

determining at higher carbon levels and hence the rate of decarburization is limited by the

supply of oxygen to the reaction sites As such the rate of oxygen flow through the bath

becomes rate determining (Fruehan 1976 Wei and Zhu 2002)) It was also noted that the rate

equations proposed by Wei and Zhu (2002) at higher carbon content did not show any noticeable

predicted change in the carbon composition at the same gas volumetric flows and time intervals

In general higher gas flow rates are required to take carbon to lower values due to oxygen flow

rate into the bath being rate limiting when carbon content is still high in the bath (Fruehan 1976

Turkdogan and Fruehan 1999 Wei and Zhu 2002) The observed trend can be attributed to the

fact that the rate equations for this model were designed for lower carbon contents in the range

of carbon content of 2 wt particularly pertinent in the refining process of stainless steels and

are significantly lower than the levels experienced in the context of this study (ie carbon

content in the range 65 ndash 75 wt ) In addition the observed trend could also be attributed to

the extent of oxygen utilization as the determination of the oxygen utilization ratio still remains

a very difficult parameter in the blowing process (Wei and Zhu 2002)

The mass and energy balance model could be manipulated to check the conditions necessary to

obtain the required carbon composition from the plant data as the model has the advantage of

predicting the conditions necessary to achieve a certain carbon composition For the collected

plant data the mass and energy balance model and the manipulation of the gas flow rates while

keeping other parameters constant resulted in achievement of the same carbon contents in the

final alloy bath obtained in the process plant data

532 Comparison of the final chromium composition based on models and plant data

The chromium content predicted from the mass and energy balance model and rate equations

were plotted against the actual plant data and results plotted in figure 57 At higher carbon

levels ie the first stage of the blow the rate equation for chromium loss was also incorporated

in the model proposed by Wei and Zhu (2002) In most cases for the first stage of the blow the

mass and energy balance model produced a better prediction than that obtained from the rate

equation based on plant data which for most parts of the blow had the lowest chromium

predicted in the bath at the end of the first stage blow for the different heats

67

Figure 57 Comparison of final Cr in the alloy bath for the first stage of the blow at high carbon content

The chromium comparison was also done between the mass and energy balance and the process

plant data for the second blowing stage and results shown in figure 58 below

Figure 58 Comparison of final Cr in the alloy bath predicted from rate equation model and plant data during the second stage of the blow

It is also important to note that the model predictions are based on equilibrium conditions being

achieved in the bath (Heikkinen 2011 Wei and Zhu 2002) However the attainment of

equilibrium conditions may be difficult to during the whole duration of the blowing stages as

there tend to be many variables changing at any given time especially in a manual controlled

63

64

65

66

67

68

69

70

71

72

73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

C

r

no of heats Rate Eqn Model

63

64

65

66

67

68

69

70

71

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

r

no of heats Cr - Model Cr-plant data

68

process where the operatorrsquos expertise has a significant impact on the process path (Heikkinen

2011) Furthermore the rate equation model proposed by Wei and Zhu (2002) was derived at

lower chromium contents of approximately 18 wt Cr whereas the application of the model in

the present case is for bath composition with greater than 65wt Cr

The discrepancy in terms of chromium composition of the bath will also have a bearing on the

accuracy in prediction of the proposed rate equation model The deviation from plant data and

model was particularly evident in the second stage of the blow (as shown in figure 58) where

the model predicted lower chromium contents than the actual plant data obtained The deviation

may be attributed to the model assumptions that all the oxygen blown into the AOD takes part in

the oxidation reactions of silicon chromium and carbon which in reality may not be the case

The mass and energy balance model as well as the rate equations (in the first blowing stage) do

not also take into account that oxygen has post combustion reactions in the converter which

might also contribute to the deviation between actual plant data and model predictions (Wei and

Zhu 2002)

533 Effect of initial temperature on the extent of decarburization

The mass and energy balance model was used to predict the effect of changing the initial bath

temperatures on the extent of decarburization The effect of initial temperature conditions were

investigated by fixing the other parameters such gas flow rates blow time interval gas ratios

and initial composition of the alloy bath The initial bath temperatures were varied between

1600degC and1780oC and results shown in Figure 59

69

Figure 59 Effect of initial temperature on the extent of decarburization at high carbon content

As shown in Figure 59 an increase in the initial bath temperature would result in the attainment

of lower final carbon content if all another critical parameters remain unchanged The trend

observed was supported by findings by Wei and Zhu (2011) based on the effects of the initial

temperature on the decarburization in a combined-blown and top-blown AOD converter Based

on the heat studied the end carbon after varying the initial temperature by +10K decreased from

0035-0034 wt while that of Cr increased from 179-1792 wt in the metal bath (Wei

and Zhu 2002 Fruehan 1976) With a decrease in of 10K the end point carbon (carbon content

at the end of the oxygen blowing process) increased to 0036 wt and the Cr decreased to

1788 wt The same analysis was done for lower carbon contents in the alloy bath (second

blowing stage) and the results also showed a decrease in the final carbon achieved with an

increase in initial temperature

However the thermodynamic feasibility as a result of the effect of initial temperature on the

decarburization process and hence the beneficial effects of high initial bath temperatures do not

take into account the need to control the starting temperatures in order to protect the refractory

materials In the actual plant operation the changes in the process temperatures are constantly

being monitored so as to avoid overheating of the AOD reactor thereby mitigating the

thermodynamic and kinetic benefits of operating at high initial bath temperatures

275

280

285

290

295

300

305

310

315

320

1600 1650 1700 1750 1800

Fin

al c

arb

on

(w

t

)

Initial temperature ( oC)

70

In figure 510 the initial alloy bath temperature was also plotted against predicted final carbon

content achievable The same trend was also observed as in figure 59 and as discussed from the

work by Wei and Zhu (2002)

Figure 510 Effect of initial temperature on end point carbon at low carbon contents

54 Effect of metal bath composition (Si Cr and C) on the distribution ratio of oxygen

The distribution ratio of oxygen among elements in an alloy bath can be presumed to be

proportional to the elementsrsquo Gibbs free energies of their oxidation reactions in the alloy bath

From the oxidation reactions of Cr Si and C shown in literature the distribution ratios of oxygen

amongst these elements was used to derive the expressions to calculate how oxygen was

distributed in the alloy bath elements (Wei and Zhu 2002 Tohge et al 1984)

The blown oxygen was seen to have different distribution ratios for Cr C and Si when

calculated As the process progresses these ratios will change with change in metal bath

temperature and composition The initial distribution ratios of blown oxygen for Cr C and Si for

the plant data are shown in the figure below Carbon had the highest ratio of blown oxygen

which means that most of the blown oxygen went towards decarburization

The distribution ratios for blown oxygen were then taken for carbon only and plotted on a graph

versus initial temperature as shown in figure 511

090

095

100

105

110

115

120

125

130

1600 1650 1700 1750 1800

End

po

int

C

Initial Temperature (oC)

71

Figure 511 Relationship between initial temperature and distribution ratios of blown oxygen for Si Cr and C

The trend showed that the distribution ratios increased with an increase in initial temperature

This is attributed to the fact that the Gibbs free energies were used for determining distribution

ratios of blown oxygen and they take into account temperature of the bath Turkdogan and

Fruehan (1999) stated that the Gibbs free energies were a function of temperature only hence

the trend observed in figure 511

A plot of oxygen distribution ratio for carbon oxidation was constructed as shown in figure 512

The trend from the graph also showed that the distribution ratio of oxygen in the carbon

oxidation also increased with increasing temperature

Figure 512 Effect of initial temperature on distribution ratios of blown oxygen for carbon

010

015

020

025

030

035

040

045

050

055

060

1600 1620 1640 1660 1680 1700 1720 1740 1760

x i

Temperature (oC) xC xCr xSi

0534

0536

0538

0540

0542

0544

0546

0548

0550

1600 1620 1640 1660 1680 1700 1720 1740 1760

x C

Temperature - degC xC Linear (xC)

72

55 Effect of initial chromium content in the bath

The effect of initial chromium content (Cr) on the extent of decarburization was also

investigated based the mass and energy balance model The basis of this investigation was to

investigate the potential effect of the interaction between the Cr and the C in the metal bath As

discussed in section 532 chromium affinity for C can potentially impact the extent on the

decarburization process In this case the effect of initial Cr content of the bath was investigated

by fixing other parameters such as initial bath temp initial metal analysis gas flow rates and

time interval for the blow The results were plotted and the trend shown in the figure 513 below

Figure 513 Effect of initial chromium on decarburization

The Figure 513 shows that with an increase in initial chromium content the end point carbon

also increased and the relationship between the Cr and C follows an almost linear trend

56 Effect of O2 partial pressure on the extent of decarburization

The oxygen partial pressures calculation was shown in section 421 The average carbon for

higher and lower carbon percent for the first and second stages respectively were obtained and

compared as a function of the average partial pressure of oxygen gas It was observed that for the

carbon content of 787 wt the partial pressure of O2 was about 451x10-13

while a carbon

content of 153 wt corresponded to O2 partial pressure of 806x10-11

The plant data showed

that the partial pressure for oxygen was lower for the first stage of blowing than the partial

pressure for oxygen at lower carbon levels in the second stage of blowing

000

050

100

150

200

250

300

350

50 52 54 56 58 60 62 64 66 68 70 72 74

Δ

C

chromium Higher initial C Lower initial C

73

The calculated values are in agreement with the studies done by Heikkinen et al (2011) where

the authors observed that the partial pressure of oxygen blown into the converter increased with

decreasing carbon content and further increased with decarburization time This phenomenon

was well observed in the second stage of blowing as shown in Figure 514 With the decrease in

silicon content due to oxidation into the slag the SiO2 activity in the slag increase and the

oxidation of carbon becomes the main oxidation occurring in the bath but once critical carbon is

reached there is an increase in chromium oxidation as well Silicon will oxidize first followed

by carbon and chromium (as silicon and carbon get depleted in the alloy bath) (Heikkinen et al

2011 Fruehan 1976 Wei and Zhu 2002)

According to Turkdogan and Fruehan (1999) as the carbon content becomes depleted and

critical carbon is reached the mass transfer of carbon to the reaction interface becomes slower

and required partial pressure to oxidize carbon increases This is illustrated in figure 514 which

shows that an increase in carbon content in the alloy bath resulted in a decrease in the oxygen

partial pressure required for carbon oxidation

Figure 514 Relationship between carbon mass vs PO2

The relationship between temperature and the partial pressure of oxygen as shown in figures

515 and 516 for carbon oxidation were analysed for first and second stages of blowing

respectively It was observed that for both the first and second stages of decarburization the

0

5E-11

1E-10

15E-10

2E-10

25E-10

05 1 15 2 25 3

PO

2 i

n t

he

allo

y b

ath

carbon content in alloy bath

74

higher the temperature the higher the oxygen partial pressure This was illustrated by Heikkinen

et al (2011) when the authors plotted graphs of RTln(PO2) against temperature in an attempt to

study how temperature played a role on partial pressure of oxygen

The figure 515 below shows the relationship between temperature and partial pressure of

oxygen for the first blowing stage

Figure 515 Relationship between temperature and partial pressure at higher carbon levels

0

5E-13

1E-12

15E-12

2E-12

1620 1630 1640 1650 1660 1670 1680 1690 1700 1710 1720

LNP

O2

Temperature - oC CLinear (C)

75

Figure 516 illustrates the same relationship as figure 515 but for second blowing stage as

discussed

Figure 516 Relationship between temperature and oxygen partial pressure at lower carbon levels

57 Effect of CO partial pressure on the extent of decarburization

The relationship between the partial pressure of carbon monoxide (CO) required for the

oxidation of carbon in the bath was investigated It was observed that the partial pressure of CO

increased with the increase in mole fraction of carbon in the bath for both the first and second

stages of the blow In general the higher the carbon content in the bath the more CO that will be

generated during the decarburization process (Heikkinen et al 2011) Figure 517 below

illustrates the relationship between molar concentrations of carbon in the alloy bath with the

partial pressure of carbon monoxide

-5E-11

0

5E-11

1E-10

15E-10

2E-10

25E-10

1680 1700 1720 1740 1760 1780 1800

Oxy

gen

par

tial

pre

ssu

re

C Linear (C)

76

Figure 517 Calculated partial pressure of CO plotted as a function of mole fraction of carbon in the bath

58 Fitting of model and plant data at lower carbon levels

The second stage of blowing commenced once the first stage was completed (ie carbon content

in the alloy bath was below 20 wt ) In the second blowing stage the oxygen and nitrogen

ratio was adjusted to 11 respectively Data collected from the plant process was used to model

the rate equation by Wei and Zhu (for lower carbon content in the alloy bath) and the mass and

energy balance model that was developed

581 Comparison of modelled and plant data in terms of carbon removal

The energy and mass balance model as well as the rate equations as derived in chapter 4 and

section 2581 ndash 2582 were utilized to investigate decarburization at lower carbon contents in

the alloy bath As already stated in chapter 3 lower carbon content referred to carbon contents

that were below 2 wt in the process under study

At lower carbon contents the mass and energy balance model was also utilized to determine the

gas flow rates necessary to achieve same plant data results At this stage gas flow ratio for

oxygen and nitrogen utilized in the converter for the second blowing stage was 11 respectively

The model was evaluated on the effectiveness of prediction of the outcome of the second

blowing stage carbon compared to the plant data results

000

020

040

060

080

100

120

140

160

180

200

050 100 150 200 250 300 350

PC

O

Carbon mole

77

The initial temperature chemical analyses and time intervals for the duration of the second

blowing stage as obtained from the process data were kept constant with the exception of the

gas flow rates that were then varied until the final carbon content compared closely to the plant

data The results of the comparison are shown in Figure 518

Figure 518 Comparison of the decarburization between modelled and plant data

As shown in Figure 518 there is a close prediction among the rate equation (mass and energy

balance and the rate equations) models and the plant data The two models under investigation

showed accurate prediction of the final carbon compared to the final carbon content obtained

from the plant data For the same flow rates chemical analyses and second stage blowing time

the extent of carbon removal in the two models accurately predicted the final carbon at the end

of the second blowing stage as observed in the comparison with actual plant results This

illustrated that both the mass and energy balance model and rate equations could accurately be

utilized as predictive tools for the second blowing stage which involves lower carbon content

59 Effect of gas flow rate on decarburization

The flow rate of oxygen directly influences the rate of decarburization with extent of

decarburization increasing with increase in oxygen flow rate when the amount of blown oxygen

is still rate limiting (ie for higher carbon contents) (Turkdogan 1980 Wei amp Zhu 2002) As

discussed in Chapter 52 plant data was analyzed for the extent of carbon removal from the alloy

bath (ΔC) for the first and second blowing stages at different oxygen flow rates The oxygen

flow rates in Nm3 were correlated to the duration of each blow to determine the amount of

060

070

080

090

100

110

120

130

140

150

160

170

180

190

200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

C

arb

on

no of heats Rate eqn Plant Data model

78

oxygen required to effect the carbon compositional change The amount of oxygen used per

interval was then plotted against the change in the carbon content during the different stages of

blowing in the AOD heats and the results are shown in Figures 519 ndash 520 for the first and

second stages respectively

It was observed that an increase of the volume of oxygen blown into the AOD resulted in an

increase in the amount of carbon removed from the alloy bath This is due to an increase in the

amount of dissolved oxygen (for carbon oxidation) per unit time resulting in increased moles of

oxygen in the alloy bath that can react with carbon to form CO gas consequently leading to an

increase in the extent of carbon removal from the alloy bath (Turkdogan and Fruehan 1999

Fruehan 1976) However a higher oxygen gas flow rate alone may not directly translate to a

higher carbon removal in a real process plant scenario as many variables can affect the amount

and rate of carbon removal from the metal bath Nevertheless the mass and energy balance

models and the rate equations developed in this study can be used to evaluate the impact of the

volumetric flow rate of oxygen on decarburization after fixing all the other parameters

The other important factor governing the maximum permissible volumetric flow rates of oxygen

blown is the rise in temperature as a result of the exothermic reaction pertinent in the AOD

refining process As discussed in Chapter 252 excessively high temperatures presents

operational challenges such as thermal degradation and the chemical dissolution by slag of the

refractory lining materials In general typical refractory materials used in the AOD reactor often

start to disintegrate at temperatures above 1750oC (Schacht 2004 Pretorius and Nunnington

2002) and in this current operation the permissible temperature increase is limited to 1780oC in

the actual plant operation the operator adjusts the volumetric gas flow rates by lowering the

oxygen flow rate while simultaneous increasing the volumetric flow rate of nitrogen in an effort

to manage the temperatures of the bath Figure 519 shows the extent of decarburization with

oxygen flow rate into the alloy bath for the first blowing stage

79

Figure 519 Effect of oxygen flow on decarburization in the first stage of blowing

The extent of decarburization with respect to oxygen flow rate was investigated and the results

shown in figure 520 below

Figure 520 Effect of oxygen flow decarburization for the second stage of the blow

510 Relationship between Chromium content and Cr activity in the metal bath

The increase in molar concentrationcontent in the alloy bath results in an increase in the activity

of chromium in the same alloy bath (Pelton and Bale 1986 Fruehan 1976 Turkdogan and

Fruehan 1999) This effect of chromium content in the alloy melt on the activity of chromium

45

5

55

6

65

7

75

350 400 450 500 550 600 650 700

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

0

05

1

15

2

25

100 150 200 250 300

Δ

C

O2 volNm3 per interval ΔC Linear (ΔC)

80

was (as investigated by these researchers) was also investigated for the process plant data

collected The chromium contents measured in the alloy bath for different heats were used to

calculate the activity of chromium to evaluate the existence of the relationship between the

molar concentration of chromium and the activity of chromium in the alloy baths Figure 521

shows the relationship between chromium content in the alloy bath and the activity of

chromium

Figure 521 Relationship of chromium content and activity of chromium in the alloy bath

511 Effect of chromium on the activity of carbon in the alloy bath

Ohtani (1956) proposed that the chromium had an effect on the activity of activity in the bath

and that the addition of chromium to Fe-Cr-C baths lowered activity of carbon The proposition

by Ohtani (1956) was also supported by work done by Richardson and Dennis (1953) In the

present study the plant data was taken for both the first and second stages of the blow and

plotted to evaluate if the chromium in the metal baths of the different heats had any effect on the

activities of carbon in the respective baths The calculated carbon activities were plotted against

the various chromium contents and the results are graphically depicted in Figures 522-523

09

091

092

093

094

095

096

097

098

099

66 665 67 675 68 685 69 695 70 705 71

γC

r

chromium γCr Linear (γCr)

81

Figure 522 Effect of chromium on the activity of carbon at high carbon contents (first stage of blowing)

The effect of chromium on carbon activity was also illustrated in figure 523 for the second

blowing stage as shown below

Figure 523 Effect of chromium content on activity of carbon at high carbon contents (second blowing stage)

As shown in Figures 522-523 the higher the chromium content of the metal bath the lower the

corresponding carbon activity The observed trend is in agreement with the work done by Ohtani

(1956) In addition the decrease in the activity of carbon with increasing chromium is

0300

0350

0400

0450

0500

0550

66 665 67 675 68 685 69 695 70 705 71

γC

chromium γC Linear (γC)

0250

0300

0350

0400

0450

0500

67 675 68 685 69 695 70 705 71 715 72

γC

Chromium γC Linear (γC)

82

significantly higher in the first stage of the blow with higher carbon content than in the second

stage of the blow corresponding to lower carbon content

According to Ohtani (1956) as decarburization continues the attractive energy between the

chromium and the carbon which governs the affinity of Cr to C tends to increase due to the fact

that the atoms of Cr and C lower each otherrsquos chemical potential In addition the temperature

also plays an important role in lowering the activity of carbon in the same way as the increase of

chromium in the bath (Ohtani 1956) This will negatively impact on refining as it becomes

increasingly difficult to decarburize as a result (Ohtani 1956)

512 Relationship between Cr2O3 content and the activity of Cr2O3 in slag

Ban-Ya (1992) stated that in a multi-component slag the activity coefficients could be described

by the quadratic formalism for regular solution Turkdogan and Fruehan (1999) went on to state

that for typical AOD slags (for stainless steel production) slag basicity is high it translated to a

higher silicon content in the steel and consequently a lower equilibrium slagmetal distribution

of chromium resulting in a higher chromium recovery in slag

From the plant data obtained a graph was plotted to investigate this relationship between

chromium oxides content in the slag and the corresponding activities of the chromium oxides

The mole fraction for Cr2O3 and the activities of Cr2O3were calculated as shown in Chapter 452

Figure 524 illustrates the relationship between the chromium oxides content is slag to the

activities of the chromium oxides

83

Figure 524 Relationship between chromium oxide and activity in slag

The chromium oxides tend exist both as Cr2+

and Cr3+

ions in slag (Ban-Ya 1992 Gasik 2013)

However the amount of Cr2+

ions in the slag depends on the total chromium oxide content in the

slag (Gasik 2013 Leuchtenmuumlller et al 2016) This means that as the chromium oxide content

in the slag increases the amount of Cr2+

decreases and Cr3+

increases (Leuchtenmuumlller et al

2016) As such for the purposes of this study it was also assumed that the predominant species

was Cr2O3 (Cr3+

ions) due to the high Cr2O3 content in the slag (Pretorius and Nunnington

2002) As shown in figure 524 the higher the mole fraction of Cr2O3 in the slag the higher the

activity of Cr2O3 A higher Cr2O3 content in slag results in the formation of spinels of MgO and

FeO which have high melting points and make slag viscous leading to higher metallic chromium

loss due to entrainment (Pretorius and Nunnington 2002 Arh and Tehovnik 2007)

The distribution of chromium oxides activity with differing temperatures of the melts for

different heats was also investigated and the results are shown in Figure 525 Though no clear in

the trend is observed for this particular set of data studies done by Xiao and Holappa (1992)

showed that the higher the bath temperature the lower the activities of chrome oxides albeit a

small effect of temperature on the activities This can be attributed to the fact that an increase in

temperature of the bath will also increase the solubility of the chrome oxides in slag thereby

reducing the activities (Arh and Tehovnik 2007) The reduction stage in the refining process

with the addition of FeSi recovers Cr2O3 that is in the slag phase Visuri et al (2012) stated that

FeSi is added as the reductant to reduce Cr2O3 and fluorspar added to reduce slag viscosity

0200

0250

0300

0350

0400

0450

0500

0550

0600

0650

0700

1000 2000 3000 4000 5000 6000 7000

a Cr2

O3

XCr2O3

84

Figure 525 Relationship between temperature and chrome oxide activity

The activities of chromium oxides in slag are also strongly influenced by the basicity (CaOSiO2

ratio) of slag (Holappa and Xiao 1992 Sano 2004 Arh and Tehovnik 2007) From the results

plotted in Figure 526 an increase in activity of chrome oxides in slag was observed with an

increase in the basicity of slag Holappa and Xiao (1992) attributed this behavior to the fact at

low basicity there is only a few oxygen ions and this result in chromium oxides having lower

activities due to a strong complexation with SiO2

Figure 526 Effect of slag basicity on activities of chrome oxides

0800

0900

1000

1100

1200

1300

1400

1500

1600

1700

1800

1660 1680 1700 1720 1740 1760 1780 1800

a Cr2

O3

Temperature - oC

020

030

040

050

060

070

080

040 060 080 100 120 140

a Cr2

O3

Basicity (CaOSiO2) aCr2O3 Linear (aCr2O3)

85

512 Nitrogen solubility in the alloy bath during decarburization

In the process under study nitrogen was used as the inert gas in the AOD converter However

nitrogen control in the alloy bath is critical to avoid precipitation of titanium nitride (TiN) and

for the production of low nitrogen bearing alloys especially for high chromium steels (Kim et al

2009) This is attributed to nitrogen solubility which increases with increase in chromium

content and will tend to decrease with increase in sulphur content (Turkdogan and Fruehan

1999 Fruehan 1992 Kim et al 2009)

In this study alloy samples were analysed for dissolved nitrogen and the results showed an

average of 12 wt nitrogen in the alloy after the reduction stage A plot of chromium content

against the logarithm of the activity of nitrogen (the logarithm function utilized to produce a

linear trend) as shown in figure 527 showed a correlation with the research by Kim et al (2009)

The graph shows a general increase in the solubility of nitrogen with increasing chromium

content This shows that the relationship between Cr by mass and lgfNCr

was independent of

the partial pressure values of nitrogen This agrees with the work done by Kim et al (2009) who

proved that nitrogen partial pressure does not have a significant impact on nitrogen solubility

The increase in chromium content in the metal bath results in an increase in nitrogen solubility in

the metal

The dissolution of nitrogen is governed by Sievertrsquos law which states that solubility of any

diatomic gas in a metal or alloy is proportional to the square root of that gasrsquos partial pressure

and the results obtained are in agreement with the work done by Kim et al (2009) based on

samples containing up to 30 wt Cr Furthermore Kim et al (2009) proposed that in Fe-Cr

alloys with up to 30 Cr by mass the Sievertrsquos law of nitrogen dissolution holds

Fruehan (1992) proposed that nitrogen dissolution in alloy bath could be reduced through

degassing blowingpurging the bath with an inert gas (such as argon) and use of special slags

(containing CaO BaO Al2O3 and TiO2) The company under study has resorted to purging with

argon during reduction stage (recovery of chromium oxide in slag with FeSi as the reductant) as

a technique of reducing nitrogen in alloy bath But the high chromium content in the alloy bath

poses a challenge as it promotes nitrogen dissolution thus the extent of nitrogen removal The

basic slags favour desulphurization so the effect of sulphur content on nitrogen removal as

discussed by Fruehan (1992) is not realized (lt 003 wt sulphur)

86

Figure 527 Relationship between Chromium and nitrogen activity

513 Financial modelling of the AOD refining process

A financial model was constructed to determine the net smelter returns for the process under

study Cost model involved costs incurred from the EAF as this had bearing on the overall costs

of the refining process as well Costs such as raw material costs (high carbon ferrochromium

alloy fines at 95 USclb burnt lime and dolomite) and consumables such as electrodes ($2

525ton of electrodes) and refractories were included The costs were expressed in $USclb of

chromium to equate to the market trend of buying and selling of ferrochromium alloys (as they

are sold according to units of chromium in the metal) Electricity was also observed to be

another big cost driver for the re-melting of high carbon ferrochromium alloy fines ranging

between R085 ndash R200kWh according to off peak and peak rate respectively according to

Eskom rates at the time of the study

The AOD costs included the cost of gases blown into the converter as well as the refractory

lining and material utilized The main cost drivers remained linings and cost of gases blown into

the converter Argon cost R783kg nitrogen R108kg oxygen R107kg and Liquid petroleum

gas (LPG) utilized for heating ladles cost R762kg From this financial model it can be

calculated per heat to investigate whether a process is economical or not The longer the AOD

process takes the more gases consumed and consequently the higher the cost of production The

-00415

-00410

-00405

-00400

-00395

-00390

-00385

-00380

-00375

-00370

-00365

-00360

66 67 68 69 70 71

logf

NC

r

chromium content -

87

selling price of the final product at the time of the study was USD$175lb and average costs

amounted to USD$135 leaving a margin of approximately USD$040lb The model was

created to augment the mass and energy balance model which can optimize the blowing process

The financial model is then used to evaluate on whether the optimization done are cost effective

financially thereby creating a tool not only for technically and scientifically evaluating a

process but also financially evaluating the economic feasibility of these technical innovations in

process optimization Appendix 17 illustrates the template constructed for cost modeling

Summary

The mass and energy balance and rate equation models were used as predictive tools for

decarburization in the production of medium carbon ferrochromium alloy for the process under

study Upon investigation it was observed that at higher carbon contents the mass and energy

balanced model results correlated to the carbon contents obtained in the plant data for the

different heats The rate equations at high carbon contents however were significantly different

from the plant data and this was attributed to the development of the rate equations which were

constructed for lower carbon contents in stainless steel production which are significantly lower

than the carbon contents under study (Wei and Zhu 2002) For lower carbon contents both the

mass and energy balance models and rate equations predictions for final carbon content were

significantly accurate with respect to plant data results

Upon analysis of activities of carbon for different heats it was observed that the activity of

carbon decreased with increasing chromium content in the alloy bath Moreover the activity of

chromium oxide in the slag decreased with increasing in bath temperature due to increased

solubility of chromium oxide The same chromium oxide activity was also observed to increase

with increase in slag basicity (Holappa and Xiao 1992)

The extent of decarburization was related to the rate of oxygen flow and significant at higher

carbon levels (defined as carbon content gt 20 wt ) compared to lower carbon contents (le 20

wt ) This was attributed to oxygen flow rate being rate limiting at higher carbon contents and

at lower carbon contents carbon mass transfer becomes rate determining (Fruehan 1976 Wei

and Zhu 2002) Nitrogen dissolution in the alloy bath was also investigated as the process under

study uses nitrogen in place of argon as the inert gas The results showed an increase in nitrogen

solubility with an increase in chromium content This was in alignment with the research done

88

by Fruehan (1992) and Kim et al (2009) that attributed nitrogen solubility to increase in

chromium and decrease in sulphur contents in the alloy bath

89

CONCLUSION

Understanding the different roles of each parameter enables the understanding of decarburization

in the AOD By first understanding the thermodynamic relationships between different elements

and oxides and their behaviour at high temperatures helps model and predict outcomes of

complex reactions per particular interval The broad objective of this study was to evaluate the

practicability of thermodynamic and parametric modeling in the refining of high carbon

ferrochromium alloy for the process under study The secondary objectives were to formulate a

mass and energy balance model and rate equations as predictive tools for the decarburization

process By assessing the current process being employed by the company under study the

objective of optimizing performance economically through understanding the process

performance and methods of improving chromium recoveries was also investigated

Oxygen is the main oxidant in the AOD and the elements in the molten metal (Si C and Cr) are

all oxidised during decarburization It was also observed that the higher the oxygen flow rate the

more moles of oxygen in the bath at a particular time interval hence the higher the

decarburization rate At higher carbon it was observed that the oxygen flow rate required for

carbon oxidation

Activities of oxides in the slag could be analysed using quadratic formalism (Ban-Ya 1993) and

it was shown from the plant data that an increase in Cr2O3 in the slag results in an increased

Cr2O3 activity Basicity of the slag also had the same effect on chromium oxides activity

resulting in an increase in chromium oxides as the slag basicity increased For the process data it

was also observed that an increase in chromium content in the metal bath resulted in a decreased

carbon activity resulting in reduced decarburization The effect was not as pronounced on the

first stage of the blowing cycle due to the higher carbon content that it was on the second

blowing stage

The initial bath temperature was investigated and observed to be quite significant on the

decarburization process The effect was more pronounced when other parameters were kept

constant and initial temperature varied over a wide range The higher the initial temperature the

lower the carbon end point for a fixed time interval This was modelled and agreed with work

done by Wei amp Zhu (2002) who also varied initial bath temperature with +- 10K and observed

90

that an increase in initial temperature by 10K resulted in a decrease in the carbon end point and

the inverse also showed same result

The rate equations proposed by Wei and Zhu (2002) were adapted to predict extent of

decarburization (as well as other oxidation reactions occurring in the refining process) and how

these rate equations compare to the actual plant data obtained for the different heats under study

However the rate equations for higher carbon content did not produce results close to the

achieved plant data since the original model was designed for lower carbon contents than those

experienced in the present study Furthermore at lower carbon contents both rate equation

proposed by Wei and Zhu (2002) and the mass and energy models developed specifically for

this study could be used to predict with a higher level of accuracy the final carbon content at the

stipulated process parameters

The initial temperature of the alloy has an impact on the end point of carbon in the process due

to improved carbon removal efficiency with increase in temperature (Heikkinen 2013

Turkdogan and Fruehan 1999) However there is only a limited increase on initial temperature

permissible in practice in order for refractory material to last for longer as the refractory bricks

start disintegrating and dissolving into the bath at bath temperatures in the excess of 1800oC

(Arh and Tehovnik 2007 Schacht 2004)

The oxygen flow rate will also impact on the rate of decarburization The higher the flow rates

the faster the carbon removal and the more the moles of oxygen available to oxidize a larger

carbon percentage in the metal bath The effect of chromium content on carbon activity becomes

more pronounced the lower the carbon content in the bath This is as a result of the affinity of

chromium on carbon which becomes more pronounced at lower carbon contents hence

negatively impacting on decarburization and increasing chromium loss to slag (Fruehan 1976

Ohtani 1956) At lower carbon contents below critical carbon content the oxidation of

chromium becomes more pronounced as the rate of decarburization is now determined by mass

transfer of carbon to reaction interfaces and not on oxygen flow rate (Wei and Zhu 2002

Turkdogan and Fruehan 1999 Fruehan 1976 Arh and Tehovnik 2007)

The parametric modelling of parameters involved in decarburization of high carbon ferrochrome

melt to produce medium carbon ferrochrome is feasible hence leading to a better control on the

91

process and better predictability of the process Through thorough understanding of

thermodynamic principles it is possible to model behaviour of the slag and alloy baths in the

converter to predict the overall outcome of the extent of decarburization for the refining process

92

RECOMMENDATIONS

The study involved the investigation of thermodynamic and parametric modeling in the refining

of high carbon ferrochromium alloys This was essential in understanding the effect of different

parameters (such as gas flow rates and ratios amongst other parameters) in the refining process

in order to attain the desired final product through the optimal control of these parameters and

their influence on refining The areas discussed below are points that require further study and

work to ensure better understanding of the decarburization process in the making of medium

carbon ferrochrome alloy

The study that was done for this particular process and plant data had its own shortcomings

There is a need for further investigation (on the interaction between the different elements in the

alloy bath under controlled conditions to investigate accuracy of the model in ideal conditions

that can be obtained in a lab set up

The physical modeling of AOD is an area for further investigation For the purpose of this study

the physical values proposed by Wei and Zhu (2002) were used in the modeling with rate

equations and may only be accurate under the conditions investigated by the researchers

Parameters such as converter size bath depth tuyere size and orientation all have impact on the

extent and rate of decarburization The determination of oxygen utilization potential is

contentious in literature with very few researchers having investigated this concept

Dimensionless constants and parameters such as bubble sizes were assumed from the work done

by Wei and Zhu (2002) and not from calculated values for the process under study The

determination of these constants still needs to be done for refining of high carbon ferrochrome as

most the researchers investigated systems for stainless steel production and very little work was

done for refining of high carbon ferrochromium alloys

The models used for the mass and energy balance model had limits as to the composition of the

Fe-C-Cr system being investigated The model used for nitrogen solubility in alloy bath used by

Kim et al (2009) was for a Fe-C system with chromium content 15 ndash 60 wt and the process

under study had chromium contents of between 65 ndash 70 wt This was further compounded by

the use of the interaction coefficients published by Sigworth and Elliot (1974) which were

investigated for dilute solutions in liquid iron with iron as the solvent and only accurate for Fe-

Cr solutions with low chromium concentrations The process under study has chromium

93

concentrations of gt 65 wt and as chromium is the main element in the alloy bath

determination of interaction parameters might need to further investigate the practicability of Fe-

C-Cr systems with chromium as the solvent The interaction coefficients were also calculated at

1600oC though there were temperature variations in the process under study

Further model development based on these issues raised is required The areas discussed below

are points that require further study and work to ensure better understanding of the

decarburization process in the making of medium carbon ferrochrome alloy

1 The determination of interaction coefficients specific to high carbon ferrochromium

alloys As already discussed the interaction coefficients used in this study were

published by Sigworth and Elliot (1974) However such interaction coefficients were

investigated for dilute solutions that have iron as the solvent thereby making them only

suitable and accurate for solutions of Fe-Cr melts containing lower concentrations of Cr

A further study to determine the interaction parameters by taking into account the high

chromium content in HCFeCr melts requires determination of interaction parameters

taking into account Cr as the solvent

2 The investigation of nitrogen solubility done in the scope of this study was based on a

model developed by Kim et al (2009) This model was designed for evaluating stainless

steel systems for chromium content up to 30 wt and partial pressures for nitrogen

004-097 atm The chromium contents involved in this study exceed the range and

hence could potentially affect accuracy of the models Therefore investigations into

nitrogen solubility at chromium contents exceeding 60 wt as the increase in chromium

content in the alloy bath increases nitrogen solubility (Fruehan 1992)

3 The chromium contents predicted by the mass and energy balance model showed

difference between the actual and the predicted final compositions for chromium and

silicon This can be attributed to the models adopted that were initially created to evaluate

Fe-Cr-C systems containing up to 30 wt (most models in literature developed for

stainless steels) chromium As a result there is need to investigate further the effects of

higher Cr content on the decarburization process of Fe-Cr-C melts particularly the

potential effect of taking chromium as the matrix instead of iron as is commonly used in

dilute Fe-Cr-C systems

4 Physical modeling was not done in this study The dimensionless constants used were

based on work done by the researchers such as Wei and Zhu (2002) but were applicable

94

for stainless steels There is need for further study to determine parameters such as

bubble sizes and tuyere configurations applicable for the current set up of the process

under study

95

REFERENCES

1 Wei JH and De-Ping Zhu DP (2002) Mathematical modelling of the argon-oxygen

decarburization refining process of stainless steel Part 1 Mathematical model of the

process Metallurgical amp Materials Transactions 33B 111-119

2 Richardson FD and Dennis WE (1953) Effect of chromium on the thermodynamic

activity of carbon in liquid iron- chromium-carbon metals Journal of Iron Steel Institute

257-263

3 Heikkinen HP and Fabritius T (2012) Modeling of the Refining Processes in the

Production of Ferrochrome amp Stainless Steel Recent Researches in Metallurgical

Engineering - From Extraction to Forming Intech publishers Shanghai Available at

httpwwwintechopencombooksrecent-researches-in-metallurgical-engineering-from-

extraction-to-formingmodelling-of-the-refining-processes-in-the-production-of-ferrochrome-

and-stainless-steel Date accessed 6112016 65-88

4 Pretorius EB and Nunnington RC (2002) Stainless steel slag fundamentals from

furnace to Tundish Iron and Steel making Journal vol 29 (2) 133 ndash 139

5 Heikkinen EP (2013) On the Role of Computational Thermodynamics in Research and

Development of AOD and processes Available at

httpccoulufi~kamaheiztktHeikkinenEP-vkpdf date accessed 6112016

6 Heikkinen EP Ikaumlheimonen T Mattila O and Fabritius T (2011) A

thermodynamic study on the oxidation of silicon carbon and chromium in the

ferrochrome converter The 6th European Oxygen Steelmaking Conference Stockholm

229 ndash 238

7 Chuang HC Hwang WS and Liu SH (2009) Effects of Basicity and FeO Content

on the Softening and Melting Temperatures of the CaO-SiO2-MgO-Al2O3 Slag System

Materials Transactions Journal Vol 50 No 6 1448 -1456

8 Goldstein M (1978) Diffusion modeling of the Carburization Process Metallurgical

Transactions Journal Vol 9A 1515 - 1525

9 Arh B and Tehovnik F (2007) The oxidation and reduction of chromium of stainless

steels in an electric arc furnace Institute of Metals and Technology Journal Vol 41(5)

203 ndash 211

10 Bhardwaj BP (2014) The complete book of Ferroalloys Niir Consultancy Services

New Delhi India p185 ndash 200